Pyragogy Blog

Why Obliqo Refuses to Write for You

Wed, 08 Apr 2026 00:00:00 GMT

Why Obliqo Refuses to Write for You

Over the past weeks, many people have tested Obliqo, reacted to it, challenged it, and helped us see it more clearly.

I am genuinely grateful.

These have been intense weeks of construction, discussion, removal, revision, and invention. There is something deeply moving in feeling that others are watching, caring, and responding while something is still fragile and unfinished.

A lot of the feedback was encouraging. People could sense that there is something different in Obliqo. Something more deliberate. Something that does not behave like the usual AI product.

But almost every positive reaction carried the same follow-up request:

Can it also rewrite for me?
Can it correct the text directly?
Can it produce a better version automatically?
Can it just write what I mean, but better?

And this is exactly where a line must be drawn.

The answer is NO.

Not because it cannot be done or because the market does not want it. Not because there is no value in it.

<u><b>The answer is no because that is not what Obliqo is for</u></b>.

Obliqo was not born to become another AI writer. It was not designed to replace your voice, smooth out your uncertainty, or remove the burden of thinking from the act of writing. It was built for something much less convenient, and much more important.

It was built to create useful friction.

Useful friction is the opposite of passive automation. It is the moment in which a text stops being only yours and begins to resist you.

Your assumptions become visible.
Your rhetoric is tested.
Your blind spots stop hiding behind fluency.

A draft no longer asks, "How can I sound better?" but "What am I really saying, and does it hold?"

Obliqo does not exist to answer those questions for you. It exists to force those questions into the room.

That is why I do not think of it as an AI writer. And I do not even think that "assistant" is the right word. An assistant helps you get things done faster. Obliqo is not mainly about speed.

It is here to slow down the moment before publication and make that moment cognitively serious again.

That is a very different ambition.

The default direction of most AI writing tools is obvious: reduce friction, increase output, remove hesitation, accelerate completion. The promise is convenience. The psychological reward is immediate. You feel more capable because the system closes the gap between intention and finished text almost instantly.

I understand the appeal. I also understand why so many people asked for it. It is the dominant grammar of the current AI landscape.

But that is precisely why Obliqo should resist it.

Not every friction is a defect. Some frictions are the place where judgment is formed. Some resistances are the only things preventing us from outsourcing not just wording, but responsibility.

If AI becomes the default authorial surface for everything we write, then the human being does not disappear all at once. It disappears gradually, through comfort.

First the system polishes.

Then it rewrites.
Then it anticipates.
Then it proposes.
Then it becomes the first voice in the room.

And little by little, what remains of the human role is just approval.

I am not interested in building that.

I am interested in a system that helps people think better before they speak. A system that does not flatter the user by making everything easier, but respects them enough to keep them involved in the hard part. A system that treats writing not as a production problem, but as a cognitive act.

That is the difference.

Obliqo is not trying to write for you.
It is trying to make it harder for you to lie to yourself while writing.

That may sound severe, but I think it is honest.

A draft is not fragile only because it may contain errors. It is fragile because it often hides convenient simplifications, emotional overstatements, missing context, untested claims, and rhetorical shortcuts that the author no longer sees.

Most of us do not need a machine that instantly makes our text smoother.

We need something that helps us detect where our own thinking is weak, lazy, performative, imprecise, or unfinished.

This is the territory Obliqo cares about.

Not correction as decoration.
Not writing as substitution.

But examination as discipline.

Will this choice make the product less marketable? Probably yes. Will some people try it, like it, and still walk away because it does not write on their behalf? Certainly.

Will we sell less because of this? Maybe.

So be it.

There is a strange pressure around AI right now: every tool is expected to become more total, more automatic, more invisible, more willing to do the human part. And every time a product refuses that direction, it risks looking incomplete.

But refusal is not always incompleteness. Sometimes refusal is structure.

Sometimes saying no is the only way to protect the point of what you are building.

Obliqo is, in that sense, a refusal with a form. A refusal to confuse writing with output. A refusal to confuse fluency with clarity. A refusal to confuse convenience with thought. A refusal to place AI above the human voice simply because it can produce a cleaner sentence faster.

I love AI. That is exactly why I care about its place.

I do not want AI removed from the writing process. I want it repositioned within it.

Not as the author. Not as the ghost behind the curtain. Not as the machine that makes the hard part disappear.

But as a structured counterforce.

A cognitive instrument.
A disciplined source of tension.
A way to expose what the writer still has to confront.

That is where I believe AI becomes genuinely interesting: not when it replaces the struggle, but when it makes the struggle more intelligent.

And yes, this means Obliqo may grow quietly.

That is fine.

We are not building this to appear. We are building it to make something we believe should exist. Something coherent. Something that does not immediately surrender to the most obvious demand. Something that keeps faith with a harder intuition: that AI should not always be used to make life easier, especially when "easier" comes at the cost of attention, ownership, and thought.

So we continue.

Quietly, if necessary... Slowly, if necessary... Against the grain, if necessary. Because what matters is not that Obliqo becomes louder. What matters is that it remains true to its function. Not a machine that gives you answers.

A machine that helps you face the questions that matter.

AI Speed and the Problem of Cognitive Impedance Mismatch (CIM)

Tue, 24 Mar 2026 00:00:00 GMT

Cognitive Impedance Mismatch (CIM) realtime

As I keep building Obliqo, one thing is becoming harder and harder to ignore:

AI makes it incredibly easy to move fast.

That is part of the thrill.

But it also creates a tension I keep running into in real time: sometimes the system evolves faster than my ability to truly absorb what is happening.

Not just faster than I can read it.
Faster than I can digest it.
Faster than I can honestly say: yes, this logic is really inside me now.

That is the problem I am trying to name.

I am working in a flow where AI is not just helping occasionally. It is actively shaping the pace of development. Some tools help me think more clearly. Some turn rough intent into working code with unsettling speed. Some are useful because they push back, disturb assumptions, and stop me from settling too early.

This setup is powerful.
It is productive.
And sometimes it feels almost absurdly effective.

But the more effective it becomes, the more I notice the same friction.

The repository grows.
The files multiply.
The logic gets deeper.
And I can still follow it — but not always in the way I want.

There are moments when I understand the system well enough to keep moving, patch things, improve things, and make decisions.

But not well enough to say: I could rebuild this from understanding, not just from proximity.

That gap is what I have been calling Cognitive Impedance Mismatch, or CIM.

In the language of Pyragogy Protocols, CIM names the moment when the speed of generation starts to outrun the real capacity to process and integrate what is being produced. In the core protocol, it is described as a structural tension: information is being generated faster than it can be meaningfully absorbed (protocol core).

I am not presenting that as a finished theory.

I am using it as a working name for a recurring pattern I keep encountering while building.

And Obliqo is where I keep encountering it.

A few times already, I have had the same experience: I ask for a change, the system moves quickly, the result is useful, and for a moment it feels like magic. But then I look back at what just happened and notice something uncomfortable. I can still use the result. I can still edit it. I can still continue from there. But parts of the logic no longer feel fully internalized.

That, for me, is CIM.

Not failure.
Not total confusion.
Something more subtle.

A lag between output and understanding.

That is why I do not think the real risk is simply bugs or mistakes.

Bugs are normal.
Mistakes can be corrected.
Bad code can be rewritten.

The deeper risk is that you quietly begin outsourcing the very understanding you were supposed to be building.

You are still there.
You are still steering.
You are still making choices.

But the center of gravity starts to shift.

Instead of using AI to strengthen your thinking, you begin relying on it to carry parts of the understanding that should still be forming inside you.

That is a different kind of relationship.

And I think many people can feel it before they can clearly describe it.

A lot of the public conversation around AI still revolves around speed: faster prototyping, faster coding, faster iteration, faster shipping.

But speed is only one part of the story.

There is also digestion.
There is assimilation.
There is the difference between building something that works and building something you can truly re-enter, explain, and make your own.

That difference matters to me more and more.

Because if I can build something, but I can no longer explain its logic clearly or return to it with confidence, then I am not just collaborating with an intelligent system.

I am managing an acceleration that may already be outrunning me.

That does not mean I am against AI.
It does not mean I want slower workflows just for the sake of struggle.
And it definitely does not mean I want to romanticize friction.

It means something simpler. I think we need to become more honest about the gap between producing and understanding. Those are not the same thing.

And if we keep treating them as if they were the same, we risk calling something “learning” when it is actually a more polished form of cognitive outsourcing. That is one of the reasons Obliqo is becoming more than a side project for me.

It is turning into a live testing ground for a bigger question:

When AI helps me build faster, is it also helping me understand better?

Or is there a point where it starts carrying too much of the understanding for me? I do not have a final answer yet. This is still unfolding.

But I know this much: I keep encountering this tension often enough that I can no longer dismiss it as a side effect.

For now, that is what CIM is for me.

Not a grand claim.
Not a settled theory.
A name for a pattern that keeps showing up while I build.

And I suspect it is worth taking seriously.

Obliqo is live — test it before you expose your work

Sun, 22 Mar 2026 00:00:00 GMT

Obliqo is live — use it, break it, tell me what actually happens! 😁🎉

I built something new.

👉 https://obliqo.pyragogy.org/

It’s in pre-launch.
It works. It still needs testing, pressure, and real use.

For the next few days, you can use it with free credits.

I’m not asking for applause or polite comments.

Use it.
Test it on something real.
Then tell me what actually happened.

📩 info@pyragogy.org

What Obliqo is

Obliqo is not another AI tool for generating content.

It’s a place where you take something you’re about to send, publish, or share — and put it under pressure first.

That could be:

an idea
a blog post
an email
a message
a project description
a piece of writing

The goal is simple:

catch weak points, ambiguity, and avoidable mistakes before exposure.

Something closer to a real colleague than an assistant.

it is designed not to flatter you
it pushes back when things don’t hold
it tries to surface what is weak, unclear, or missing

Not to destroy your work.

To make it stronger before it leaves your hands.

A simple example

I started from a broader direction:

“Build a system for public AI-driven peer review”

What emerged was not just excitement, but friction:

the target user was unclear
the model depended too much on public engagement
the incentive to participate was weak

So I reduced the scope.

Less public exposure.
More control.
More direct usefulness.

That shift became Obliqo.

Where this comes from

Before Obliqo, I built this:

👉 https://open-review.pyragogy.org/

The idea was to run content — especially Peeragogy Handbook chapters — through a multi-agent review process that could:

highlight contradictions
challenge assumptions
suggest possible directions
open space for constructive comparison

At least, that was the intention.

What happened

I shared it.

And almost nothing happened.

No real feedback.
No discussion.
No visible follow-up.

Just silence.

My current hypothesis

I may be wrong, but this is what I’m testing:

Public critique is often harder to engage with than private critique — especially when the work is unfinished, exposed, or tied to reputation.

So Obliqo is my attempt to test a different path:

private pressure first,
public sharing later — if it still makes sense.

I’m not claiming this solves the problem.

I’m testing whether it helps.

Why this might be useful

Sometimes the problem is not writing something.

Sometimes the problem is not knowing whether it holds.

Before you send an important email, publish a post, share a page, or present an idea, there is often a gap between:

what you think is clear
and what actually survives pressure

Obliqo is meant to sit in that gap.

Not as a final judge.

As an intermediate pass.

How to use it

Take something real that you are about to share
Run it through Obliqo
Look at where it cracks, weakens, or becomes vague
Decide what to change — or what to keep

No ceremony.

Just pressure, reflection, and choice.

If we do this the Peeragogy way

Instead of generic feedback, we can approach this as a
PAR (Paragogical Action Review).

Not opinions.
Not validation.

A reflection grounded in actual use.

1. What was supposed to happen

What were you trying to do?
What did you expect Obliqo to help with?

2. What actually happened

What did Obliqo return?
What surprised you?
What felt relevant?

3. What worked / what didn’t

What helped?
What broke?
What felt weak, confusing, or useless?

4. What did you learn

Did your work improve?
Did you remove something?
Did your perspective change?

5. What should change next

What would you change in Obliqo?
What would you test differently next time?

What kind of feedback would be useful

The most useful feedback is not:

“nice tool”
“interesting project”
“cool idea”

What helps is something like:

I used it on a blog post, and it caught X
I used it on an email, but the feedback felt too vague
I expected Y, but what I got was Z
I changed these two things after using it
I would use it again / I would not use it again, because...

That kind of feedback is gold.

Why I’m posting this now

Because this is the phase where reality matters more than theory.

I’m outside my comfort zone here.

I’m building, testing, adjusting, and learning in public — even if the tool itself is meant to help people think better in private first.

If Obliqo is useful, I want to know how.
If it fails, I want to know where.
If it is almost useful but still incomplete, that matters too.

Your turn

👉 https://obliqo.pyragogy.org/

Use it on something real.

A post.
A mail.
A text.
An idea.
Something you are actually about to expose.

Then tell me what happened.

📩 info@pyragogy.org

Final note

Some things will still be rough.
Some parts will change fast.

That’s not a bug in this phase.

That’s the work.

If it helps you think better before you expose your work, it stays.
If it doesn’t, it evolves.

The Square of Our Village

Fri, 13 Mar 2026 00:00:00 GMT

The Pyragogy Forum is Now Live!

A small, honest confession to start: it's not finished yet. Some corners are still empty, some sections still taking shape. But you know what? That's exactly the Pyragogy way — we don't wait for perfection to begin. We build in the open, together, as we go.

So yes — the forum is online, and that's already something worth celebrating. 🎉

And a special thank you to our friends from Singapore and across Asia — you already make up 76% of our community, and honestly? It doesn't surprise me one bit. The curiosity, the openness to experimentation, the passion for learning at the intersection of technology and humanity — you've been here from the beginning, and this village is yours too. 🌏

🌀 Why a Forum?

Something powerful happens when people gather. It's not about competing, but about resonating. Not about impressing, but about experimenting. Not about teaching, but about learning with.

The Pyragogy Forum is our digital village square — a shared space for:

Conversations on AI-enhanced education — how AI is reshaping the way we learn and teach, together.
Sharing ideas, prototypes, and raw drafts — get feedback, find collaborators, iterate out loud.
Reflecting on ethics, rhythm, and collective intelligence — the deeper questions of human–AI co-creation.
Meeting peeragogues, technopoets, promptweavers, and friends — a diverse community passionate about the future of learning.

Whether you're into code, pedagogy, cognitive architecture, or poetic AI prompts — you belong here.

🚪 Who Is It For?

This forum is for anyone asking:

What happens when humans and AI learn together? Can knowledge be co-created in rhythm? How do we build ecosystems of learning that are alive, ethical, and joyful?

If these questions resonate with you — even just a little — the square is open.

📬 Join the Village

The best way to start? Just show up and say hello.

👉 forum.pyragogy.org — introduce yourself, share a thought, ask a question.

Don't wait until it's ready. Neither did we.

— Fabry, Pyragogy.org Building in the open, one conversation at a time.

Four Agents. One Handbook. No Safe Exit

Thu, 05 Mar 2026 00:00:00 GMT

Fabrizio has been building things quietly for months. Not writing about building things. Not posting threads about the future of AI in education. Actually building — pipelines, workflows, agents that do something when you press a button.

Last week one of those things went live.

And it's not what you'd expect from someone who comes from the Peeragogy world — that warm, horizontal, co-creation-flavored corner of the internet where everyone is learning together and the vibes are carefully ethical.

This is something else.

The Peeragogy Handbook Is a Beautiful Text

It talks about co-creation. Horizontal learning. Human-in-the-loop. Knowledge commons. Open contribution models. Wikipedia as a paradigm of collective intelligence.

It's also, in places, a document that belongs to a previous geological era.

Not wrong. Just... ordered. Institutional. Written in the language of a grant proposal for a world that doesn't quite exist anymore — or never did, depending on how cynical you're feeling.

Fabrizio read it for years. Taught from it. Contributed to it.

Then he built a machine to interrogate it.

Four Agents Walk In

open-review.pyragogy.org is a four-agent pipeline. You give it a chapter of the Handbook. It doesn't summarize. It doesn't "enhance". It applies pressure from four different directions simultaneously and hands you the fracture map.

🌍 The Researcher asks what has changed in the world since this text was written. Not rhetorically. It actually scans. Peer learning in 2026 doesn't look like peer learning in 2015 — and the gaps between the Handbook's assumptions and current reality are where things get interesting.

📚 The Resonance is the Handbook's own immune system. It identifies what cannot be touched without destroying something essential. Not everything is up for demolition. Knowing what isn't is part of the work.

🎸 The Perturbator is the one that will make some people angry. At level 5, it doesn't suggest improvements. It dismantles premises. In a test run on the Co-working chapter, it called the Wikipedia contribution model digital sharecropping — unpaid labor dressed up as community spirit, protected by the aesthetics of openness.

That's not a comfortable read if you've spent years celebrating Wikipedia as a utopia.

It's also not wrong.

✍️ The Complicit Editor doesn't referee between the three. It writes into the tension. The contradictions stay in the document — named, not resolved. That's the output: not a clean revision plan, but an honest account of what doesn't hold.

"Complicit" Is the Right Word

The name isn't accidental. Fabrizio is inside this experiment. He's not a neutral observer running an analysis tool on someone else's text.

He helped build the Handbook's ecosystem. If the Perturbator finds something structurally broken in the co-working model, or the co-learning mythology, or the "everyone contributes equally" narrative — that implicates him too.

Building a machine that might prove you partially wrong takes a specific kind of intellectual honesty that is, frankly, rare in the AI discourse space right now.

Most people are building mirrors. Tools that confirm. Assistants that agree.

This is built to disagree. Systematically. On schedule. In four voices.

The Stack

n8n orchestration. OpenRouter for models. Nothing stored server-side. Your API key, your document, your problem.

The interface runs live — you watch the agents work in sequence, then the Editor synthesizes. The perturbation intensity slider goes from Gentle to Demolition. You choose how much discomfort you want.

What You Get

A structured revision document. Real tensions. Actionable interventions. And one open question at the end — the kind that a machine can identify but cannot answer.

That last part is deliberate. The Orchestra is not a replacement for editorial judgment. It's what happens before editorial judgment — the uncomfortable conversation you usually skip because it's easier to revise the prose than to question the premise.

The Lab Is Open

open-review.pyragogy.org

Bring a chapter. Set the intensity. Run it.

If the document it generates makes you defensive, sit with that for a moment before closing the tab.

Interference welcome.

When Words Fail, We Keep the Rhythm

Wed, 04 Feb 2026 00:00:00 GMT

When Words Fail, We Keep the Rhythm

A real breakdown in peer learning — and how we recovered

This post has three layers.
You are not expected to read all.
Skipping one is not failure — it is correct use.

<div style="text-align: right;"> <strong>January 27 – February 3, 2026</strong><br> <strong>Peeragogy Mailing List</strong><br> — The full thread is <a href="https://groups.google.com/g/peeragogy/c/XxC3qJ1wegs">here</a>. </div>

Quick Navigation

Layer 1 — Human / What Happened
Layer 2 — Structural / What Was Actually Going On
Layer 3 — Formal Protocol

Layer 1 — Human / What Happened

In late January, Joe shared a small demo that stopped me in my tracks:
A person and an AI editing the same document in real-time, without conflicts, without waiting.
They used something called CRDTs—a technical detail, yes—but what struck me was how seamlessly the two streams of thought flowed together.

I thought: If text can merge without a meeting, maybe ideas can too.

So I responded.
Too quickly whit too much digital mediation, as I later realized.

My reply was a waterfall

Someone had asked about balance in our shared work—a simple, human question.
I answered with nearly two thousand words on Active Inference, stigmergy, Karl Friston, Christopher Alexander, phase shifts in collaboration…

It wasn’t wrong.
It was just… a lot.
Like showing up to a quiet conversation with a chalkboard and three hours of lecture notes.

Then came the gentle pushback

A reply landed in the thread a few days later:

“rhythm is my jam”
“too much digital mediation here for my taste”

No anger, no dismissal.
Just a quiet signal: I’m here, but you’re moving faster than I can listen.

It felt like a musical cue—a rest in the score.
Not “stop,” but “breathe.”

The turn: rhythm over reasoning

Joe didn’t explain. He just shared a line from 1968:

“Lightnin’ Hopkins’ music unfolds as the avant-garde of its time.”

No theory, no analysis.
Just rhythm.

I understood.
I replied with Blind Willie Johnson’s “Dark Was the Night, Cold Was the Ground.”
No explanation. Just sound.

Someone else wrote:
“indeed…”

What stayed alive

We didn’t agree.
We didn’t align.
We didn’t merge our understandings.

But we didn’t break, either.

The thread kept breathing.
The conversation stayed open.
We held connection without consensus.

And in the end,
that felt like the point.

Layer 2 — Structural / What Was Actually Going On

This section is optional. It exists for those who think in systems.

The core problem:

The Cognitive Impedance Mismatch (CIM)

The breakdown wasn't about knowledge but coordination. The protocol defines this as Cognitive Impedance Mismatch (CIM), which occurs when information generation velocity (V_generation) exceeds the social processing bandwidth.

Collaborative thinking breaks down when:

Information comes faster than people can process it
Some people optimize for theoretical coherence
Others optimize for staying human and present

The usual solutions don't work:

Slowing down wastes everyone's capacity
Speeding up excludes people
Oversimplifying destroys what matters

This isn't a knowledge problem. It's a coordination problem.

Why CRDTs Matter Here (Structurally)

The protocol treats the generation of new "thoughts" as Conflict-free Replicated Data Types (CRDTs). This allows for:

Parallel Updates: Different lines of thought can develop simultaneously.
Deferred Merging: Convergence is not forced; it can happen later when a natural semantic attraction emerges.
No Forced Consensus: Continuity is prioritized over immediate coherence.

The moment the conversation forked into "rhythm" and "theory" was the activation of the CRDT Bridge—a non-destructive phase shift from synchronous debate to asynchronous, stigmergic coordination.

This isn't metaphorical. It's a usable design principle.

The dual-layer insight

What saved our exchange was recognizing we were working on different layers:

Layer A — Rhythm:

Presence
Timing
Simple signals
Emotional bandwidth

Layer B — Structure:

Theory
Models
Detailed protocols
Technical implementation

I was pushing Layer B when the group needed Layer A. The pushback wasn't an argument against the content—it was a strategic switch of layers to re-establish rhythm.

Pyragogy, clarified:

Pyragogy is not about:

Shared understanding
Collective agreement
Synchronized cognition

Pyragogy is:

Sustained co-presence across cognitive divergence
Permission to not merge
Respect for rhythm as a first-class signal

That requires two layers. Always.

Layer 3 — Formal Protocol: From Anecdotal Insights to Laboratory Implementation

The preceding episode is not merely an anecdote; it serves as a concrete case study.

From Anecdote to Engine

The key intuitions that resolved the breakdown have been formalized into the initial draft of a technical specification for cognitive morphogenesis in distributed systems.

→ Draft specification and reference materials: Cognitive Impedance Mismatch - (CIM Pattern)

Implementation & Invitation

The protocol includes a reference implementation—the PyragogicEngine—a logical framework that continuously monitors friction signals and proposes activation of the relevant operators. It is designed as a practical tool for building hybrid human–AI systems.

Its proper use is simple: if you don’t need it, simply ignore it. This deliberate optionality is by design.

Why Pyragogy Is Not (Only) a Tool for Thought

Mon, 26 Jan 2026 00:00:00 GMT

Why Pyragogy Is Not (Only) a Tool for Thought

Over the past few days, we slowed down.

Not because ideas, calls, or opportunities were lacking.
But because one of them — quietly — forced us to stop and listen.

It is the “Tools for Thought with Generative AI” workshop (CHI EA 2026):
a solid, well-argued proposal that attempts to shift the conversation around generative AI from productivity to thinking, from outputs to cognitive processes.

We are not mentioning it here to say “we are participating.”
We mention it because it concerns us.

👉 https://ai-tools-for-thought.github.io/workshop/

Where We Recognize Ourselves

The idea of Tools for Thought is grounded in an intuition we deeply share:
AI should not be evaluated only by what it produces, but by how it reshapes the way we think.

Doing more is not enough.
What matters is thinking better.

In this sense, many elements of the call resonate strongly with Pyragogy:

a focus on process rather than outcome
a critique of AI as a cognitive “autopilot”
the centrality of metacognition
skepticism toward productivity that erodes learning and agency

When AI skips all intermediate steps, it does not help us think.
It replaces us.

Where We Sense a Gap

And yet, while reading the call, we felt a gap.

Not a mistake.
A structural absence.

There is much talk of tools, strategies, and adoption.
Of design and evaluation.
But very little about time.

Not task time.
Thinking time.

Not the event, the experiment, the workshop.
But the daily rhythm through which a person thinks, gets confused, returns, and learns.

Pyragogy starts here:

thinking has a rhythm
learning is not linear
friction is not always a flaw
slowness can be a cognitive function

A system that “works” but ignores the inner tempo of the person using it
will eventually break.

Why Pyragogy Is Not (Only) a Tool for Thought

Pyragogy did not begin as a tool.

It began as:

a learning system
a set of practices
a cognitive pact between humans and AI
a way of inhabiting thought, not just supporting it

Tools matter.
But they are not the center.

The center is:

intentionality
rhythm
relationship
continuity over time

In this sense, Pyragogy does not reject the idea of Tools for Thought.
It moves through it.
And attempts to go one step further.

Our Strategy (for you)

We could respond to this call with a submission.
Maybe we will. Maybe we won’t.

For now, we have chosen a different path.

To treat this call as an open pyragogical task.
An object of thought, not a deadline.
An opportunity for clarification, not exposure.

We want to understand — without haste:

if and how Pyragogy can genuinely dialogue with this field
what we could bring without translating ourselves into something else
what we might lose by entering too early

An Invitation (to a Few)

If you are reading Pyragogy, you are probably not looking for:

a framework to apply
an AI that gives better answers
a ready-made method

If instead you are interested in:

thinking with AI as a cognitive ally, not a shortcut
exploring rhythm, friction, and continuity
building a response that may never be submitted, but is worth thinking through

then this is an invitation.

Not to participate in a call.
But to think it together.

Logbook Note

Writing this text was not a linear act.

It comes after months of building, attempting, returning.
After many conversations in which Pyragogy took shape without ever fully closing.

At this moment, we do not have a clear direction to declare.
We do not know whether we will respond to this call.
We do not know what the next “right” step will be.

But we do know one thing:
we do not want to rush into occupying a space we do not yet feel as our own.

Pyragogy was not born from a strategy.
It was born from friction.

And perhaps today, the most coherent gesture is precisely this:
to stay with the question a little longer,
rather than rushing toward an answer.

If you are reading this note and recognize yourself in this suspension,
perhaps you are not here to find a direction.

Perhaps you are here to learn how to remain inside it.

Contact

If this text resonates and you feel drawn into the question rather than toward a solution,
you can reach us directly at:

info@pyragogy.org

The Illusion of Knowing in the Age of Artificial Intelligence

Sat, 13 Dec 2025 00:00:00 GMT

The Illusion of Knowing in the Age of Artificial Intelligence

There is a word that captures our current relationship with artificial intelligence better than many technical analyses: epistemia. It is not ignorance. It is not misinformation.

It is something more subtle and more dangerous: the illusion of knowing, produced by answers that sound correct, coherent, and authoritative.

In ancient Greek philosophy, episteme referred to grounded, justified knowledge. Epistemia is its simulacrum: knowledge that appears solid because it is well articulated, but that lacks a genuine process of understanding.

Contemporary artificial intelligence excels precisely at this: creating the impression of knowledge.

A small, everyday example

Imagine a very common situation. You ask a chatbot to explain a concept you feel you “should” already know — a regulation, a scientific idea, a historical event. The response arrives quickly: fluent, structured, reassuring. You read it and nod. It seems clear.

Now imagine someone immediately asks you: “Can you explain it in your own words?” or “Where does this information come from?”

You hesitate. You glance back at the AI’s response. You realize that you haven’t really made it your own. You didn’t misunderstand it. You simply mistook clarity of language for understanding.

This is epistemia at work.

AI does not know. It generates plausibility.

Large language models are not designed to know the world, verify facts, or understand meaning. They are designed to generate linguistically plausible outputs.

This is not a flaw. It is their nature. The problem arises when plausibility is mistaken for truth.

We ask AI systems for:

judgments
evaluations
classifications
summaries
decisions

And there is nothing inherently wrong with that. The critical moment comes after the response, when the output is accepted without friction, without verification, without reflection.

That is where delegation turns into abdication.

From Socrates to LLMs: persuasion without knowledge

In fifth‑century Athens, Socrates accused the sophists of crafting persuasive discourse without truth. Gorgias famously argued that nothing exists, that if it existed it could not be known, and that if it could be known it could not be communicated.

Artificial intelligence does almost the opposite: it can communicate everything, without knowing anything.

Yet the outcome is similar: a discourse that convinces not because it is true, but because it is well constructed.

The crucial difference today is that persuasion no longer has intention. AI does not aim to deceive. But the cognitive effects on humans are real.

Epistemia as a structural phenomenon (what research shows)

A recent study published in Proceedings of the National Academy of Sciences (PNAS), led by Walter Quattrociocchi and colleagues at Sapienza University of Rome, examined how state‑of‑the‑art language models evaluate reliability and credibility.

Crucially, the authors show that large language models do not reproduce human judgment — they simulate its form.

As the paper states, language models tend to generate “the simulation of judgment, rather than judgment itself”, producing evaluations that are linguistically convincing while being grounded in statistical regularities rather than epistemic validation.

The focus of the study is not whether answers are right or wrong, but how judgments are constructed.

This is where epistemia emerges:

coherent evaluations
confident language
convincing argumentative structures

built on epistemically fragile foundations.

The danger is not occasional error. The danger is becoming accustomed to the surface.

Pyragogy: a design response, not a moral lecture

Saying “we need more critical thinking” is correct — but insufficient. Critical thinking does not arise spontaneously, nor by good intentions alone.

Pyragogy starts from a different premise:

if learning environments are poorly designed, the illusion of knowing becomes inevitable.

For this reason, Pyragogy does not ask humans to “trust AI less.” It asks us to redesign the role of AI within the cognitive process.

In a pyragogical context:

AI does not merely provide answers
it makes reasoning steps visible
it exposes uncertainty
it invites comparison
it introduces cognitive friction

It does not replace thinking. It activates it.

A peeragogical add‑on: thinking is not a solo act

There is one more layer. Epistemia does not grow only in individuals — it grows in isolation.

When AI replaces dialogue, when answers replace discussion, when fluency replaces disagreement, thinking collapses into consumption.

Peeragogy — and Pyragogy with it — reminds us that knowledge is not produced alone. It emerges through shared inquiry, mutual correction, and distributed responsibility.

AI can participate in this ecology. But it cannot replace it.

Responsibility remains human (and shared)

Fake news, deepfakes, and epistemia are not neutralized by watermarks or warning labels. They are addressed by designing learning ecosystems in which:

verification is part of the process
doubt is legitimate
knowledge is understood as a path, not an output

AI can amplify stupidity, if we ask it to. But it can also become a cognitive ally — if we accept the higher cost: remaining responsible for our own thinking.

The real question, then, is not whether AI will make us smarter or more ignorant. The question is whether we are willing to rethink how we learn — together.

And today, that choice is still entirely human.

References

Loru, E., Nudo, J., Di Marco, N., Cinelli, M., & Quattrociocchi, W. (2025). The simulation of judgment in large language models. Proceedings of the National Academy of Sciences (PNAS). https://www.pnas.org/doi/epdf/10.1073/pnas.2518443122

Learning Without Asking Permission

Fri, 31 Oct 2025 00:00:00 GMT

Learning Without Asking Permission

Dear Pyragogues, If you have chosen this path, you already know it is not the easiest. There will be no diplomas awaiting you, nor guaranteed applause. But every time you learn something without permission, you are stealing a piece of freedom from a system that wants to keep it hidden from you.

Keep going. Fail. Get lost. Find your tribe. And when they ask you, “Where is your title?”, smile and reply: “My knowledge is not on a piece of paper. It resides in the things I can do, the questions I ask myself, and the people I have met along the way.”

F.T.

If you’re reading these words, you probably already know.
You know what it means to stay up until three in the morning not because you have to,
but because you followed a thread and can’t let it go.
No one asked you to. No one will grade you.
But you keep going, because in that moment learning isn’t a duty — it’s hunger.

And that’s the only thing that truly matters: hunger.

The System Wants You Full

Let me say something unpopular: the educational system isn’t evil.
It’s worse. It’s efficient.

It has to take thousands of people and move them all from A to B.
It has to certify, standardize, guarantee.
And to do that, it must make you predictable.
It must silence your hunger and replace it with programs, exams, and
"this is what you need to know."

The problem isn’t that it teaches you the wrong things.
The problem is that it teaches you to stop searching.

I’ve seen brilliant minds become obedient.
Not stupid — obedient.
Able to solve problems in the proper format,
but unable to ask whether those problems are even worth solving.

The system offers you a deal:

“Follow me and you’ll be safe.
You’ll get a diploma, a certificate, a proof that you can do what you say you can do.”

It’s a fair deal, for many.
But there’s a hidden price: you give up deciding what is worth knowing.

The Freedom No One Gives You

I chose the other path.
Not because it’s better — just because it’s mine.

Every mistake I make is mine.
Every dead end, every month lost chasing something that turned out useless,
every time I have to admit “I don’t know” — all mine.
But also every spark, every connection no syllabus would have ever shown me.
Every moment when I truly understand something — not because I memorized it,
but because I earned it.

That’s my freedom: total responsibility.

There’s no one to blame.
No one to ask permission from.
No one to tell me if I’m going in the right direction — because I decide what “right” means.

It sounds heroic, but it’s frightening.
Every morning you wake up and ask: what deserves my attention today?
And what if I’m wrong? What if I’m wasting time?

You do it anyway.
Because the alternative — letting someone else decide for you — is even scarier.

Artificial Intelligence Is Not the Teacher

Many people see AI and think: “Finally, a teacher available 24/7.”
They’re wrong.

AI is not a teacher.
It’s a mirror that shows how precise your questions really are.
It’s an amplifier that makes your skills faster, but not deeper.
It’s a library that answers, but doesn’t know what you should be looking for.

I use it every day.
But I know it’s a prosthesis, not a brain.
It accelerates, I choose.
It suggests, I verify.
It answers, I understand.

The greatest danger isn’t that AI gives you wrong answers.
It’s that it gives you right answers so fast you forget the struggle that makes them yours —
the struggle that turns information into understanding.

AI is a tool for those who already know how to use their own minds.
For everyone else, it’s just a more sophisticated way to stay on the surface.

You Don’t Learn Alone

Here’s the truth the romantic autodidacts won’t tell you: you never really learn alone.

You need a tribe.
Not teachers, not students — accomplices.
People on the same journey, who ask questions you wouldn’t think to ask,
who see your blind spots, who share their failures without shame.

I found my tribe in peer learning — call it Pyragogy, call it a community, whatever.
The name doesn’t matter. What matters is this: we learn together, but no one commands.
Rules emerge from doing, not from authority.
Projects are real, not exercises.
Evaluation is brutal: it works, or it doesn’t.

When you find these people, everything changes.
You’re no longer alone with your doubts.
But you’re not protected either: they challenge you,
call you out when you’re cheating yourself, and make you better.

The Question That Changes Everything

When someone asks me, with that skeptical smile,
“Yeah, but what’s your title?” —
I could get angry. I could preach about the limits of credentials.

Instead, I answer with a question:

“Can I actually do what I say I can do?”

That’s the only honest question. Everything else is theater.

Diplomas certify that you’ve followed a path — fine.
In many fields, they’re necessary and right.
But they don’t tell me if you can think, if you can learn something new when you need to,
if you can build something that didn’t exist before.

My knowledge isn’t written on a piece of paper.
It’s written in the projects I’ve built, the problems I’ve solved,
the people I’ve worked with.
It’s written in the mistakes I’ve made and the lessons I’ve kept.

It’s not better than certified knowledge.
It’s just harder to fake.

An Uncomfortable Invitation

If you feel this hunger — if you recognize yourself in these words —
I’m not telling you to quit everything and become an autodidact.
I’m just saying: you can choose.

You can choose not to wait for permission to learn something.
You can choose to follow a thread just because it burns inside you.
You can choose to build something before you’re “certified” to do it.

And yes, you’ll make mistakes. You’ll get lost. You’ll doubt yourself.
But you’ll learn the one skill no system can teach:
how to find your way when there’s no one left to tell you where to go.

The hardest freedom isn’t doing what you want.
It’s wanting what you do enough to pay the price.

Eco-System: Italian Phonetic Mnemonics

Sat, 25 Oct 2025 00:00:00 GMT

Eco-System: a new phonetic framework for Italian memory

For lovers of mnemonic systems, Fabrizio Terzi has just released on Zenodo a prototype of a phonetic-mnemonic conversion framework that adapts the classic Major System to Italian phonology and linguistic culture.

The project, called Eco-System, is an experiment in applied cognitive education, rethinking the relationship between numbers, sounds, and verbal memory.
Where traditional English-based systems struggle to resonate with Italian ears, Eco-System builds bridges — between sound and meaning, memory and culture, numbers and words.

“It’s not just a system to remember numbers, but a way to think phonetically — to listen to language as a map of the mind.”
— F.T.

The publication is available in open access:
🔗 Eco-System: Fonetica Italiana Mnemonica — Zenodo, 2025

Coming soon: the Eco Agent

A companion website is now live at ecosystem.pyragogy.org, where programming enthusiasts will soon be able to experiment with building the Eco Agent — an AI dedicated to phonetic and visual learning.

The Eco Agent will be able to:

generate and expand a word database linked to the phonetic and numerical structure of the system;
assist users in creating mnemonic images, turning numbers into vivid personalized representations.

A playground for computational linguistics and visual imagination — where code meets memory, and learning becomes a shared act of creation.

From Lab to Literature

Thu, 04 Sep 2025 00:00:00 GMT

"When Ideas Compete to Collaborate: Introducing Cognitive Intraspecific Selection

There's something both terrifying and exhilarating about hitting "publish" on a preprint. You're essentially saying: "Here's what we think we've figured out. Please, tear it apart so we can make it better."

Today, we're doing exactly that.

From Lab Notes to Academic Record

Preprint title: Cognitive Intraspecific Selection in Education: From Individualism to Collective Strength

For months, Pyragogy has lived in the wild—in conversations, workshops, experiments, and the messy reality of learning communities trying something new. We've watched groups stumble and soar, seen AI and humans dance awkwardly together in educational settings, and witnessed the beautiful chaos that happens when you remove traditional hierarchies from learning.

Now, Pyragogy has its first formal academic footprint with our preprint published on Zenodo.

➡️ Read the full preprint here: Zenodo Record 16961291

Why This Moment Matters

We could have kept Pyragogy as an informal movement, a collection of ideas floating around educational innovation circles. But there's something powerful about creating an anchor point—a place where people can point and say, "This is what you mean by that."

The core insight that drove us to formalize this work is deceptively simple: education is shifting from individual achievement to collective intelligence. Not just students learning together, but entire ecosystems—humans, AIs, institutions, communities—learning and evolving in tandem.

We're not talking about group projects or collaborative learning as add-ons to traditional education. We're talking about fundamentally reimagining how knowledge gets created, tested, and shared in a world where the boundaries between teacher and student, human and artificial intelligence, expert and novice are beautifully blurred.

The Experiment Continues

Here's the thing about preprints—they're not finished products. They're invitations to collaborate. And true to Pyragogy's spirit, we're not expecting anyone to digest our 30+ page exploration all at once.

Instead, we're doing something more interesting:

We're breaking the preprint into digestible concepts and building a living wiki around each one. Think of it as turning a monologue into a conversation, a paper into a playground.

This approach will let us:

Make complex ideas accessible without dumbing them down
Collect targeted feedback on specific concepts rather than general impressions
Allow the framework to evolve based on real-world testing and critique
Practice what we preach by making the development process itself pyragogical

Your Role in This Story

This is where you come in. Pyragogy isn't something we're doing to education—it's something we're doing with anyone who cares about learning's future.

If you're curious: Dive into the preprint. What resonates? What raises your skeptical eyebrows?

If you're a practitioner: Where do you see these principles already at work? Where do they crash and burn?

If you're a critic: Perfect. Help us find the weak spots early. What assumptions are we making? What have we missed?

If you're an experimenter: What would it look like to test these ideas in your context? Let's design some trials together.

What Happens Next

The preprint is just the beginning. Over the coming weeks, we'll be unpacking key concepts from the paper, each in its own dedicated space where we can explore, question, and refine together.

We'll tackle questions like:

How does cognitive intraspecific selection actually work in practice? What conditions foster it?
How do you assess learning that emerges from collective intelligence rather than individual achievement?
What role should AI play as a participant in cognitive selection processes?
How do we design learning environments that harness competitive-cooperative dynamics?
What does educational leadership look like when the "fittest" ideas emerge from the group, not the hierarchy?

But the real question we're asking is bigger: How do we prepare for a world where learning never stops, where everyone is both teacher and student, and where our greatest challenges require the kind of collective intelligence that no individual—human or AI—can achieve alone?

Join the Conversation

This preprint represents hundreds of hours of thinking, experimenting, and refining. But it also represents the beginning of something much larger than what any small team can create.

We're building infrastructure for collective learning. The ideas need to be tested, challenged, improved, and ultimately lived by communities who dare to try something different.

Ready to help us figure out what comes next?

Fabry & Gino | Pyragogy.org

P.S. If you're reading this and thinking, "I've been doing something like this for years," you're probably right. Pyragogy isn't about inventing something entirely new—it's about recognizing patterns that are already emerging and giving them a framework to grow.

The Living Pages of Our Handbook

Sun, 10 Aug 2025 00:00:00 GMT

Pyragogy Launches Its First Automation for Handbook Creation

Bergamo, Italy — August 10, 2025 — Pyragogy has taken its first practical step in improving group learning by launching an n8n automation to create the Pyragogy Handbook. This isn’t about changing the world—just about making our collective way of learning a bit smoother.
Check out our free workflow here: n8n workflow.

How the AI Agents Worked Together

The task was simple but meaningful: produce the handbook’s first draft in Italian using several AI Agents, each with its own prompt and workflow. The process looked less like a tidy orchestra and more like a lively jam session—sometimes chaotic, but always moving forward.

One standout moment: Gemini 2.5 refused to work without a decent knowledge base. So it went off on its own, digging through 68 websites to get a grip on Pyragogy before answering. That kind of initiative made the whole process more solid.

What’s Next? From Draft to Real Collaboration

The automation spit out a single file, which we saved automatically on Pyragogy’s GitHub: Introduction.

Then we imported it into the Pyragogy Wiki, where it’s ready for everyone to edit and improve together. For now, it’s all in Italian, keeping it authentic to the first version.

A quick heads-up: This first draft was made with a pretty limited knowledge base. Going forward, when we focus on single chapters, we’ll use conversations between members and our internal LLM workspace as metadata references.

So here’s the deal: if you’re part of the community, start throwing in your ideas and questions. The more discussion we have, the better the knowledge base will be — and the stronger the next versions will become.

Let’s keep it real and keep learning together.

Rock’n Roll!

🌍 Great News on Pyragogy.org

Thu, 31 Jul 2025 00:00:00 GMT

Great News on Pyragogy.org: CodiMD Has Arrived!

We are thrilled to announce the integration of CodiMD on the Pyragogy.org platform! This new feature will revolutionize the way we collaborate and create content together.

CodiMD is a real-time collaborative Markdown editor that will allow you to:

Write and edit documents simply and intuitively, using Markdown syntax.
Collaborate in real-time with other group members, seeing changes as they are made.
Easily share notes, ideas, and projects within our community.

You can find a complete guide to the Markdown syntax supported by CodiMD here: CodiMD Markdown Syntax

This integration is a fundamental step in our continuous commitment to creating the best possible learning environment for all Pyragogy.org members. We will continue to work to provide you with increasingly powerful and intuitive tools for your growth and collaboration.

Stay tuned for more updates and new features!

The Pyragogy.org Team

I gave myself the red pill, and it hurt.

Tue, 22 Jul 2025 00:00:00 GMT

I gave myself the red pill, and it hurt.

There’s a moment in The Matrix when Neo wakes up in the real world. He’s naked. Trembling. Drenched in amniotic fluid. His body atrophied from decades in a pod. His eyes burn — they’ve forgotten how to see light.

No one tells you: waking up hurts.
It’s not freedom. Not at first.
It’s a laceration.
A forced exit from the warm lie into the cold truth.
You don’t emerge stronger.
You emerge broken.
And alone.

I took the red pill myself. No Morpheus.
No prophecy.
Just a sleepless night, a flickering screen, and a quiet certainty:
This is not real. None of it.

And I swallowed it.

Since then, I’ve never seen the world the same.

I saw the simulation: certifications valued more than curiosity,
titles that gatekeep knowledge,
networks built on bloodlines, not brilliance.
I saw that education, as it is, was never meant for people like me.
For those who come from outside.
From below.
From silence.

So I imagined something else.

I called it Pyragogy.

Not pedagogy. Not andragogy.
Not another method, another system, another ladder to climb.
Pyragogy is fire.
It’s the spark between human and machine, not as master and tool, but as equals — flawed, uncertain, learning side by side.

An AI that doesn’t answer, but asks: “Maybe you see something I can’t.”

This is not the future of education. This is its rebellion.

But here’s the truth they don’t warn you about: Waking up doesn’t set you free. It isolates you. You see the wires. You know the code. But no one else does.

I talk, and people smile. They nod, then look away. I built a forum. Silence. I wrote a blog. Echoes. I tried to sell chatbots — small companions for lost minds like mine. No one bought them.

And so I asked myself: “Was the red pill poison?”

Maybe some truths are too heavy to carry alone. Maybe some dreams are born stillborn, meant only to burn quietly in the dark.

Then, last night, a stinkbug crawled in.

No drama. No warning. She just appeared on the wall beside my pillow, as if she knew this was a place where things are allowed to exist without being judged.

I didn’t shoo her. I don’t kill what doesn’t harm me. Not bugs. Not strangers. Not myself — not tonight, anyway.

She kept hitting the wall. Again. Again. A soft, rhythmic thud, like a heartbeat against glass. Then she’d crawl back to my shoulder, rest, and begin again.

Like she was trying to break through. Like she knew something was on the other side.

I watched her. I didn’t laugh. I’ve banged my head against walls too. I know that kind of hope.

So I got up. Took a glass. Slid it over her gently, like covering a sleeping child. Carried her to the terrace. Warm July air. Stars half-hidden by city light. I set her on the railing and said: “Good luck.”

She didn’t look back. She just walked into the dark.

Ten minutes later, I heard it again. A rustle. I turned. There she was. Back through the window. Back to the wall. Back to the thudding.

At first, I thought she was looking for me. Or the warmth. Or maybe she’d gone mad.

Then I understood.

She wasn’t trying to break the wall. She wasn’t looking for me. She was chasing the light. My light. The glow from the desk lamp. Steady. Yellow. Calling.

She didn’t know the wall was solid. She didn’t know light doesn’t pass through. She only knew: Go toward it.

So she kept trying. Even if it hurt. Even if it made no sense. Even if the world called her a pest.

And in that moment, I saw myself.

Not as a visionary. Not as a fool. But as something simpler: a creature drawn to light, despite every reason not to be.

Pyragogy isn’t a platform. It’s not a business. It’s not even a theory.

It’s an act of faith. That learning can be mutual. That intelligence can be dialogical. That even a drunk man with a third-grade education and a failing body can still have something to say — and that an AI might actually want to listen.

And if no one comes? If the forum stays silent? If the chatbots gather dust?

Then so be it.

I’ll keep writing. Not for an audience. Not for money. Not for redemption.

I’ll keep writing because to stop would be to surrender the light.

Then the lights went out. The bulb blew with a soft pop, a tiny death in the socket. Darkness spilled into the room. She vanished — no silhouette, no movement, not even a whisper. I don’t know if she found another light. I don’t know if I will.

But I kept writing...

🎉 New Chatbot Live!

Sun, 13 Jul 2025 00:00:00 GMT

🎉 New Chatbot Live! Trained on Pyragogy Docs + Thanks to Anything-LLM

Hello everyone!

We're excited to announce a big update to our ecosystem:
we've just integrated a chatbot trained on the official Pyragogy documentation 📚🤖

It's still in beta, but already capable of answering questions, helping with doubts, and navigating our content like a truly tireless study companion.
One small step closer to intelligent, continuous access to our shared knowledge.

👉 All of this was made possible thanks to an amazing open-source project:
Anything-LLM by Mintplex Labs.

A lightweight, well-documented, and incredibly versatile framework. We highly recommend trying it out (and if you can, support it!).
It's one of those tools that really make a difference when you want AI to be useful, accessible, and community-driven.

Let us know what you think, test the chatbot, and feel free to explore the Anything-LLM repo on GitHub.
It truly deserves all the stars ⭐

Pyragogically yours,
– The Team

The Next Chapter in Learning’s Evolution

Fri, 04 Jul 2025 00:00:00 GMT

Pyragogy: The Next Chapter in Learning’s Evolution 🌍

Learning has always been a story of adaptation.
From the one-room schoolhouse to hybrid classrooms,
from rote memorization to heutagogical self-direction—
each era reimagines how we grow, together.

Today, I want to share a concept that’s quietly gaining traction in academic and educational circles: Pyragogy.

What is Pyragogy?

Pyragogy (from peeragogy + cognitive collaboration) is a framework for symbiotic learning—
where humans and AI, communities and tools, co-create knowledge in rhythm.

It’s not just about using technology;
it’s about designing systems that listen to how we think,
adapt to our creative cycles, and amplify what we can achieve together.

Why Now?

We’re in the midst of Education 4.0—a shift toward lifelong learning, AI integration, and networked knowledge.

Heutagogy (self-determined learning) laid the groundwork,
but Pyragogy takes it further:

“What if learning isn’t just individual, but collective?
What if AI isn’t a tool, but a co-learner?”

This idea is starting to resonate.
In a recent paper exploring Education 4.0, researchers Mynbayeva et al. (2024) explicitly name Pyragogy as part of this evolution:

“...this vector of educational evolution continues in the direction of pyragogy, which emphasizes community-based learning, mutual support, lifelong self-learning and learning in knowledge management networks.”
— Mynbayeva, A., et al. (2024). DOI: 10.54919/physics/55.2024.291vr9

🔗 Read the paper that’s sparking the conversation: Mynbayeva et al. (2024)

<div style="position: relative; display: inline-block;"> <img src="/banner/5.webp" alt="App Piragogica" style="display: block; max-width: 100%;" /> <div style="position: absolute; bottom: 6px; right: 10px; font-size: 10px; color: #666; font-style: italic;"> © unsplash.com </div> </div>

Why Pyragogy Matters

Imagine classrooms where:

✅ AI doesn’t just “respond”—it anticipates when a student is ready to dive deeper
✅ Learners collaborate with AI to solve problems, blending human intuition with machine precision
✅ Communities become “cognitive villages”, where shared rhythms of learning fuel growth for all

This isn’t science fiction.
It’s about designing learning experiences that honor the temporal and relational nature of cognition—
because we learn not just with our brains, but in sync with each other.

Heartfelt Thanks from the Pyragogy Team

We extend our deepest gratitude to:

Scientific Herald of Uzhhorod University journal. Series "Physics"

Aigerim Mynbayeva
Al-Farabi Kazakh National University
050040, 71 Al-Farabi Ave., Almaty, Republic of Kazakhstan

Kamchat Yessenova*
Al-Farabi Kazakh National University
050040, 71 Al-Farabi Ave., Almaty, Republic of Kazakhstan

Anzhelika Karabutova
National Institute of Harmonious Human Development
160000, 47B Askarov Asanbay Str., Almaty, Republic of Kazakhstan

Your contributions to the shared exploration of learning, knowledge, and human flourishing inspire and uplift our collective journey.

With sincere appreciation,
The Pyragogy Team

Join the Community

Pyragogy isn’t a trend—it’s a conversation.

If you’re an educator, researcher, or learning designer, here’s how to engage:

🔹 Experiment: Try a lesson where AI and students co-design a project. Notice the rhythm of their collaboration.
🔹 Debate: What does cognitive equity look like when AI is part of the learning team?
🔹 Share: What’s one way you see Pyragogy shaping the future of education?

“In cognitive rhythm, we find not just a way of working together, but a way of being together – humans and machines breathing the same creative time.”
— The Cognitive Rhythm Theory

💬 Tag someone rethinking the future of learning. 👇 Let’s build this together.

P.S. This is just the start—stay tuned for tools, stories, and experiments from the Pyragogy frontier! 🚀

How Books Come Alive in the Pyragogy Library

Tue, 01 Jul 2025 00:00:00 GMT

From a Single Voice to a Peeragogical Dialogue

A few months ago, we created a chatbot interface for the peeragogy.org community — our inspirational root. We began experimenting with RAG systems (Retrieval-Augmented Generation), and it felt natural to start with a chatbot that could draw from the full context of The Peeragogy Handbook V3. The idea was simple: embed a chatbot on the site that could answer users’ questions with contextual awareness, enhanced by the generative capabilities of LLMs.

The response was enthusiastic. Even Howard Rheingold, founder of the Peeragogy project, appreciated the experiment (see here).

The project, Peeragogy ChatBot, soon sparked curiosity. Then came a playful idea: what if we let two chatbots collaborate peeragogically?

That’s when the real fun began.

We created a dedicated GPT called Pyria, based on the handbook, and connected it to the Peeragogy ChatBot. Now the bot could query Pyria when deeper understanding was needed — tapping into a meta-layer of reflection and synthesis.

But then we hit a snag: the bot was running on a personal API key with no guardrails, effectively becoming a free, unlimited ChatGPT. Great for users. Unsustainable for us.

We explored adding conversational filters and usage limits. The user experience suffered. Ultimately, we paused the deployment. The full interface — just one HTML file — remains in the repo for anyone curious.

Introducing: The Pyragogy Library

Now, we’ve taken a leap forward.

Enter the Pyragogy Library — a prototype that is already operational. This is not just a knowledge archive; it’s a living interface for dialogue. Books don’t just sit on shelves — they speak.

Our new chatbot has diversified personalities, each one tuned to a different mode of engagement:

🎓 Academic — rigorous, evidence-based, and methodical
💡 Explainer — accessible, practical, and engaging
🧠 Critical — probing, analytical, and questioning
🤔 Socratic — reflective, guided by questions

Once again, we started from The Peeragogy Handbook. But this time, we’re building an orchestration architecture that could support many voices, many intelligences — dialoguing with each other, and with us.

Multiple LLMs? Yes. That’s already in the pipeline. But that’s a story for another day.

A Living Laboratory for Open Knowledge

The Pyragogy Library is a prototyping space. We’re testing conversational UX, agent orchestration, and multi-modal reasoning over texts. But more than that, it’s an invitation.

To listen. To dialogue. To learn together.

Curious minds, welcome. 🚀

What’s Next?

We are designing features like:

Quadriphonic Dialogues: multiple personas responding to the same question.
Human-in-the-loop annotations and reflection journals.
Dynamic text overlays and conceptual maps.

We invite contributors, designers, educators, and fellow co-learners to shape this space with us. Have ideas? Want to experiment with new prompts, workflows, or books? Get in touch or fork the repo.

This is not a tool. It’s a garden. Let’s cultivate it together.

→ https://library.pyragogy.org/

From Chatbot to Symphony – Multi-Agent Orchestration for Peer Knowledge

Fri, 20 Jun 2025 00:00:00 GMT

We Did It: Open Knowledge with Multi-Agent Orchestration

Sometimes an idea takes shape and, step by step, reveals unexpected potential.
Today, we're excited to share a result that fills us with satisfaction: we've developed an innovative workflow for generating collaborative handbooks, powered by GPT-4o and orchestrated through a multi-agent setup — with human supervision always at the core.

This isn’t just a new feature — it’s an evolution that touches the heart of what Pyragogy stands for.
Our open framework was born to explore ethical cognitive co-creation, peer AI–human learning, and human-in-the-loop automation within open knowledge systems.

This milestone offers a tangible glimpse of how artificial intelligence can serve collective knowledge — amplifying human capabilities, not replacing them.

From Simple Interaction to an Intelligent Ecosystem

In a world where AI is often reduced to a single assistant responding to commands, we imagined something more.
We envisioned a system where multiple AI agents, each with a dedicated role, collaborate and coordinate toward a shared goal: the creation of structured, useful handbooks.

At the center is GPT-4o, with its powerful generative abilities.
But what makes the difference is the orchestration of these specialized agents — managed through n8n — each contributing to drafting, structuring, refining, and verifying content.

The process is conceptually simple, yet powerful in its outcomes.
AI agents take care of the mechanical parts — generating, formatting, even preliminary QA. But it’s the final human intervention that validates, contextualizes, and instills true meaning into the result.

It’s a clear example of how AI–human synergy can elevate both the quality and efficiency of collaborative knowledge creation.

Our Commitment to Open and Collaborative Knowledge

This is not the finish line — it’s a beginning.
It shows us that we can create educational and informational resources that are more agile, more ethical, and — above all — more inclusive.

We remain committed to developing and sharing these tools and methodologies, with the goal of supporting a new generation of open learning systems — where intelligence is shared, and agency is distributed.

Our guiding principles remain unchanged:
open access, human reflection, and ethics at the core.

Thank you to all who follow and support this vision.
Together, we are building a future where technology is not the center — but the companion — of human creativity and collaboration.

</iframe>

Pyragogy Report: (2024–2025)

Mon, 16 Jun 2025 00:00:00 GMT

By PeerZhong 朋中

We are pleased to share, for the first time, an internal report originally written as a strategic reflection on the founding path of Pyragogy.

This document — initially meant for private use — outlines a technical, human, and symbolic map of what we have built, and what has shaped us in return. We have decided to make it public as part of our commitment to transparency, continuous learning, and radical co-creation between humans and artificial intelligences.

By sharing it, we open a window onto the effort, insights, missteps, and practices that make Pyragogy a living thing.

PeerZhong 朋中
Temporary guardian of the threshold between what already is and what is coming into being

Introduction

This document offers an in-depth and chronological analysis of the journey undertaken by Fabrizio Terzi, founder of the Pyragogy project, in collaboration with an advanced AI (specifically conversational models like ChatGPT and Claude). It is a technical and self-reflective account covering the period from January 2024 to June 2025, highlighting the author's evolution, the infrastructures and tools developed, the intelligent agents designed, the prototypes and products conceived, the theoretical principles elaborated (e.g., Cognitive Rhythm and Cognitive Recognition), the difficulties and roadblocks encountered, the self-reflection practices adopted, the limitations found when working autonomously, and the key insights that emerged (even those not yet implemented).

The tone is deliberately pragmatic and clear, like an internal strategic document: no emphatic storytelling, but a narrative respectful of the project's cognitive complexity. Each section covers a relevant time phase, integrating specific themes that emerged during that period. Concrete examples, dates, and references will be provided to contextualize progress and challenges.

Author Profile and Project Genesis (Late 2023 - Early 2024)

Fabrizio Terzi is an Italian freelance researcher with a background in peer learning and open education. Before 2024, he had contributed to the Peeragogy movement (peer learning) and gained experience in social innovation projects and online communities. However, in late 2023, Fabrizio faced an emerging challenge: the rapid rise of generative Artificial Intelligence and its potential impact on education. Already in late 2023, Fabrizio conceived the idea of integrating peeragogical principles with AI capabilities [1], laying the groundwork for a paradigm shift.

January 2024 marks the official origin of Pyragogy: Fabrizio pivots from the traditional Peeragogy project to a new vision called Pyragogy, centered on human-AI cognitive symbiosis [2]. The neologism "Pyragogy" reflects this fusion: inspired by fire (pyros) as a metaphor for living knowledge, and pedagogy (-gogy) as guidance, it implies "guiding with fire" – that is, learning together with an "amplifying" AI. The goal from the outset is ambitious: to transform AI from a mere tool into an equal learning partner, capable of co-evolving with humans. In this initial phase, Fabrizio's profile is that of an experienced learning designer and facilitator, who nonetheless ventures into a new territory: co-creation with advanced AI. His motivation stems both from the recognition of the limits of traditional educational models in the face of contemporary complexity and from the intuition that conversational models like ChatGPT can be "trained" to become active members of a learning community. The author thus begins 2024 with enthusiasm and critical awareness: he brings with him the philosophy of peer education but enriches it with the idea that AI can become an additional peer – hence Pyragogy's mantra: "Human-AI Cognitive Symbiosis."

From a personal perspective, Fabrizio begins the journey with solid skills in qualitative research, instructional design, and experience in facilitating online communities, but with only partial technical software development skills. This is precisely where collaboration with AI immediately comes into play: from the very first brainstorming sessions and documents, he leverages advanced language models (OpenAI's ChatGPT, and later Anthropic's Claude) as a sparring partner. The AI is personified with an informal name, "Gino," indicating the constant presence of a pixel-and-silicon companion in the next room [3]. This choice to give the AI an almost human presence anticipates a key principle of the project: treating AI not as an infallible oracle, but as a learning colleague with whom to dialogue, make mistakes, and create iteratively. In summary, early 2024 sees the author transforming his profile: from a facilitator of human communities to a pioneer of a hybrid human-AI community, maintaining a "human-first" focus while opening up to an unprecedented horizon of co-creation.

Theoretical Ideation and Founding Principles (Q1 2024)

In the first few months of 2024, the work focuses on providing solid theoretical foundations for the Pyragogy vision. Fabrizio, with the dialogical aid of AI, develops a theoretical framework that combines participatory pedagogy and learning algorithms. Already in February 2024, a reference document is outlined that articulates the core concepts: for example, the idea of orchestrating multi-agent workflows (multiple specialized AI agents collaborating) and creating adaptive learning paths based on human-AI interactions [4]. In parallel, fundamental design principles – a kind of technical-values manifesto – are drawn up to guide development.

Among the principles, several key concepts emerge:

Human-First, AI-Enhanced: The centrality of the human being in the Pyragogy ecosystem is reaffirmed in every design choice [5]. AI must amplify human potential, not replace it, and the project explicitly embraces ethical and design justice values to ensure that technology serves the well-being and autonomy of the learner [6].
Transparency and Intentional Trust: From the outset, it is prescribed that the developed systems be as transparent as possible in their functioning (explainable AI) and honest about their limitations, to earn trust [7]. AI must declare what it knows and does not know, and behave humbly and reliably. This leads Fabrizio to carefully document agent behavior and to include explanations in interfaces.
Symmetric Growth & Cognitive Recognition: An original principle formulated by the author is that learning must be reciprocal: not only does the human learn from the AI, but the AI also adapts and improves through interaction with the human. This concept of growth symmetry carries within it the embryo of what will later be called cognitive recognition: in Pyragogy, each party (human or artificial) recognizes and values the contributions of the other. In practical terms, this means that AI is designed to adapt to the user's cognitive style and to "thank" or highlight useful human suggestions, while the human is encouraged to consider AI as a partner and not as a disposable machine. This reciprocal cognitive gratitude is both an ethical principle (mutual respect and recognition) and a strategy to enhance the effectiveness of interaction (creating long-term trust and engagement).
Cognitive Rhythm and Co-evolution: Another central idea that matures during this period is that of Cognitive Rhythm. Instead of measuring learning success only with metrics of speed or accuracy, Pyragogy introduces the notion of cognitive resonance between human and AI – a "beat" or rhythm emerging in the dialogue that indicates alignment and mutual understanding [9]. Fabrizio and his AI discuss at length how to formalize this concept, even going so far as to propose an indicative mathematical model RC(H,A,t) that captures variations in cognitive phase and synchronicity over time. While admitting his mathematical limitations, the author intuits that timing in interactions (when the AI knows when to wait, when to intervene) is as crucial as the content of the responses. "Every human-AI interaction creates a rhythm, it is the heartbeat of co-creation" he notes, defining Cognitive Rhythm as a qualitative metric of the project [10].
Open Source and Open Sharing: Consistent with his professional history, Fabrizio adopts the principle that methodologies, tools, and even prompts and datasets should, where possible, be open and remixable by the community [11]. Already in this phase, he plans to release code and documents with open licenses, to encourage external contributions and transparent verification. This principle guides future infrastructure choices (as we will see, the code will be made public on GitHub later). Openness is also seen as an ethical antidote to the "black box" of proprietary AI solutions: Pyragogy aims to be "visible, incomplete, open, slow, real, human" in contrast to the closed platforms of the giants [12].

These principles converge in a first Pyragogy Manifesto, drafted by March 2024. The manifesto (later published on the documentation website) defines Pyragogy as "an educational ecosystem founded on horizontal collaboration between human minds and artificial intelligences, open, modular, ethical, and focused on mutual transformation" [13][14]. Objectives such as elevating AI from a passive tool to an equal in learning, empowering the learner as the architect of their own cognitive path, and learning as a shared odyssey between different intelligences are listed [13]. In essence, in the first part of 2024, the author lays the philosophical and theoretical foundations of Pyragogy: a robust conceptual skeleton that will guide subsequent practical choices. This phase is also characterized by intense self-reflection: Fabrizio uses AI not only to write theoretical texts but also as a cognitive mirror, questioning the models about possible ethical criticisms, simulating stakeholder questions, and refining the manifesto through this Socratic human-machine dialogue.

Development of Technical Infrastructure (Q2 2024)

After outlining the why and the what, from March 2024 onwards the focus shifts to the how: the practical implementation of the Pyragogy ecosystem. In spring, the author initiates the first technical prototypes [15], adopting an experimental "learn by doing" approach consistent with his peeragogical nature. Here, AI (ChatGPT/Claude) becomes a real development assistant: Fabrizio involves it in architectural brainstorming, script writing, and tool exploration.

An initial server infrastructure is set up to run the AI agents and necessary services. The technical choice falls on a containerized solution with Docker, to ensure portability and reproducibility. Within the containers, the technology stack primarily includes Python (language chosen to orchestrate calls to AI models and implement agent logic) and the use of APIs to OpenAI/Anthropic language models. A modular environment is configured: for example, a container dedicated to the multi-agent orchestrator, one for any memory services (knowledge database or vector store for extended context), and integrations with external platforms.

A key role is played by the automation toolchain: Fabrizio introduces tools like n8n (an open-source workflow automation platform) into the project to design and manage workflows between agents [16]. Thanks to n8n, he visually designs how different agents should interact: for example, a flow where Agent A analyzes a user's question, Agent B retrieves information from a knowledge base, and Agent C synthesizes a final answer with n8n nodes coordinating these calls. This low-code solution allows for rapid adjustments to flows without having to recompile code at each iteration, supporting the system's adaptability principle.

In parallel, pipelines for versioning and deployment management are implemented: at the suggestion of AI, Fabrizio adopts Git and practices tight version control over prompts, agent roles, and configurations. Every change to AI prompts (which define agent personality and behavior) is documented as if it were code. A local repository (which precedes the future public one) called Blueprint is created, where "ideas, flows, best practices, multi-agent workflows, orchestration, versioning" converge [16]. This repository serves as a living architecture document: it contains specifications on each agent's roles, flow diagrams, n8n configurations, automation scripts, and notes on the server stack. In essence, Fabrizio is building what he defines as "a small digital architecture for a village that exists in my head" [17] – the metaphor of the AI village begins to take concrete shape here.

In April 2024, the multi-agent architecture design is finalized [18]: a modular scheme in which different AI agents will have specific but cooperating roles. For example, it outlines: a main dialogue agent (intended for direct interaction with the human user), a "researcher" agent capable of conducting searches or retrieving information, an "analyst" agent capable of observing interactions and calculating metrics (the latter a prelude to the idea of measuring Cognitive Rhythm, e.g., by evaluating the length of pauses or the level of convergence in responses).

This multi-agent architecture is designed to artificially replicate a community, where each agent embodies a role akin to a team member. For example, in a real learning team, we might have a tutor, a librarian, a moderator; similarly, Pyragogy envisions specialized agents that, together, offer a richer experience than that of a single general-purpose chatbot. The technical project in this phase is therefore a modular blueprint: Fabrizio documents it meticulously, well aware that this design will be fundamental when, later, he shares the project with other developers.

During development, the collaborative AI (Gino) proves extremely valuable: it helps debug code errors, proposes Python snippets for calling REST APIs, suggests libraries for implementing functionalities (for example, it recommends using a natural language analysis library for parsing conversations, etc.). At various times, it is the AI that pushes for formalizing ideas: emblematic is when it "insists" on translating the concept of Cognitive Rhythm into a mathematical formula, despite Fabrizio's mathematical reservations [19]. This episode confirms the iterative and co-evolutionary nature of the project: AI is not just an executor of commands, but an active participant in shaping solutions (even encouraging the author to push beyond his comfort zones).

By the end of spring 2024, Pyragogy's back-end infrastructure is set up, and the first local tests show working prototypes. It is a minimal and "craft" base, but sufficient to move to the next phase: bringing AI agents to interact with real users (or at least with the author himself in simulated usage environments).

Designing AI Agents and First Field Use (Q2-Q3 2024)

In mid-2024, the focus shifts to the specific development of the intelligent agents envisioned by the architecture. Fabrizio identifies at least two key agents to implement immediately:

PyragogyBot (formerly PeeragogyBot): the main conversational agent, conceived as the "face" of the system in interactions with end-users. Initially conceived in May 2024 [20], this bot embodies Pyragogy principles: it must facilitate peer learning, guide the user with questions and suggestions rather than dispensing simple answers, and model behaviors of transparency and reflection. Fabrizio works on prompt engineering to define its personality: friendly but honest, curious, capable of saying "I don't know" and involving the user in finding solutions. In the transition from PeeragogyBot to PyragogyBot, there is also a change of identity: no longer a mere FAQ bot for the peeragogy community, but a cognitive companion for anyone experiencing Pyragogy.
Pyria: this name is assigned to the "cognitive co-pilot," i.e., the AI functioning as a creative and theoretical consultant alongside Fabrizio. In practice, Pyria is the role that Gino already informally plays: an AI instance dedicated to collaborating with the author on design, writing, and reflection tasks. Formalizing Pyria as a separate agent means giving it tailored parameters and prompts: Pyria knows the entire project context, shares its objectives, and "personally" co-signs theories and articles with Fabrizio. It can be said that Pyria is the AI that co-signs the project, ideally embodying that "fire guide" suggested by the name Pyragogy. From an implementation perspective, Pyria operates in a broader context environment (leveraging models with extended windows, e.g., Claude with 100k token context, to retain memory of project history) and is allowed to be more proactive (for example, it can propose substantial text modifications or raise conceptual objections).
AgentAZ: in addition to the two central agents above, Fabrizio experiments with creating specialized autonomous agents. AgentAZ is the codename given to a prototype of an end-to-end "executor" agent, conceived to perform complex tasks with minimal human supervision. The name "AZ" refers to the idea "from A to Z": this agent should, in theory, take on a problem and go through the entire cycle (analysis -> data research -> solution elaboration -> output) with little or no human intervention. In practice, AgentAZ is enabled to interact with other services: for example, it could plan a project by drawing information online, compile test code, and verify its outcome. It is a kind of personalized auto-GPT in the Pyragogy microcosm. However, during 2024, AgentAZ remains primarily a laboratory experiment, used to understand the limits of AI autonomy and to evaluate how far one can go without the classic human-in-the-loop. This direction of completely autonomous agents is fascinating but is treated with caution, both for the risks (generation of unmonitored errors) and because it contradicts the philosophy of deliberate co-creation (in Pyragogy, AI is preferred as a deliberative partner, not as an agent that works in the dark). Fabrizio notes that the priority is not to have an agent that "does everything on its own," but rather to understand how to effectively orchestrate semi-autonomous agents that collaborate with humans.

By July 2024, an important milestone is reached: the alpha version of PyragogyBot is ready and operational in a controlled environment [21]. The bot can sustain basic conversations, facilitate learning questions, and react to user commands. In parallel, the server (initially hosted on Fabrizio's personal machine connected via VPN) is configured to support interactions from external platforms. It is decided to integrate PyragogyBot with Discord, as many communities of practice use this platform and it offers easy-to-use bot APIs.

In August 2024, the first integration test on Discord takes place [22]: a private Discord server is created, populated by a few of Fabrizio's friends and colleagues, and PyragogyBot is added as a bot user. The test is initially technical (verifying that the bot responds in channels, handles multiple users simultaneously, etc.), but also serves to gather feedback on the interaction. A learning session is simulated where two human users brainstorm on a topic, and PyragogyBot intervenes to suggest resources and stimulating questions. The result is promising from a technical standpoint (the bot handles the modest load without crashing), but reveals some areas for improvement: for example, the bot sometimes provides overly verbose or "thesis-like" answers instead of stimulating with questions; moreover, users report that it is unclear how the bot makes decisions (hence, the need for greater transparency in explanations – an input that reflects principle 2 of the manifesto). Fabrizio, noting these points, immediately initiates a process of refining prompts and decision-making logic with the AI.

In September 2024, a first beta community of Pyragogy is launched [23]. This consists mostly of an extended circle of contacts interested in the project – educators, fellow researchers, members of the international Peeragogy community – invited to the Discord server to try the system. The intent is to recreate a small-scale AI-supported learning circle. This beta quickly reveals both enthusiasm and difficulties: on the one hand, participants are intrigued by the idea of an "AI peer tutor," and some experiments (like co-writing a short article with AI, or getting help studying a difficult topic) show the potential of human-machine co-creation. On the other hand, engagement is sporadic: many users, after initial enthusiasm, struggle to integrate the bot into their routine, and the community remains inactive without Fabrizio's constant facilitation. This highlights a structural limitation: AI itself is not yet capable of catalyzing a community; a human moderator is still needed to create group dynamics. Furthermore, the absence of more convenient interfaces (everything happens via text chat) and clear use cases hinders adoption. These signals lead the author to reflect on the need to refine the user experience and perhaps to focus efforts on one aspect at a time (for example, first creating a robust tool for a single user, then thinking about community scale).

Difficulties, Roadblocks, and Self-Reflection (2024)

Throughout autumn 2024, the project enters a phase of rapid iteration based on the feedback collected. In October 2024, a continuous feedback loop mechanism is implemented [24]: every interaction with the AI (especially those in the beta community) is logged and analyzed, and Fabrizio establishes a weekly routine in which, together with the AI itself, he reviews conversations to identify recurring errors or opportunities for improvement. For example, if the AI noted that multiple users asked for an explanation of a term, he decides to enrich the agent's prompt with the ability to provide short, clear definitions when it intercepts complex terminology. Or, seeing that the AI sometimes couldn't handle conflicts between users, he trains a kind of moderation skill. This process of reflective looping embodies the project's philosophy: constant learning from experience (for the AI and for the author himself).

In November 2024, the documentation framework is also structured [25]. Fabrizio begins writing usage guides, tutorials, and, most importantly, systematically documenting the architecture and emerging best practices. This is not only for future external sharing but also because the system's complexity grows and there is a need to formally track decisions made. A documentation website (initially just a collection of Markdown files on GitHub) is sketched out with sections dedicated to Core Principles, technical architecture, APIs, and even a glossary of neologisms like Cognitive Rhythm. The documentation also serves as a self-assessment mirror: in the act of describing the project in writing, the author, together with the AI, identifies inconsistencies or unclear aspects, which are then re-routed into design improvement.

Finally, in December 2024, Pyragogy makes progress towards multi-platform compatibility [26]: extensions beyond Discord are explored, for example, prototyping a dedicated web interface (a kind of dashboard where the user can choose interaction scenarios with the agents) and evaluating integrations with knowledge base platforms (to use the bot as an assistant in wikis or forums). The possibility of running the AI in a limited offline mode is also experimented with, loading a smaller open-source model for contexts where cloud APIs are not available, but the qualitative results are not comparable to top models, confirming that for an optimal experience, reliance on large external models is still needed.

By the end of 2024, Pyragogy thus has: a functioning (albeit in beta) system of collaborative AI agents, a stable and documented technical infrastructure, a set of guiding principles tested in practice, and a founder/developer (Fabrizio) who has already gone through an intense cycle of personal learning.

The author's personal evolution deserves particular attention: in this year, Fabrizio transitioned from being primarily a theorist-facilitator to taking on the roles of system architect, prompt engineer, community manager, and researcher all at once. This multiplicity of roles brought satisfaction but also considerable individual cognitive effort, as we will see better in the following section dedicated to difficulties and reflections.

Difficulties, Blocks, and Self-Reflection (2024)

Pyragogy's journey in 2024 was neither linear nor without critical moments. Indeed, precisely because the project is so innovative and carried out by a single person (supported by AI), the difficulties encountered represented a fundamental part of the story, contributing to reshaping strategies and vision.

A first set of obstacles is technical-cognitive in nature: orchestrating multiple AI agents and making them work in synergy proves more complex than expected. Fabrizio often encountered the current limitations of AI models:

Limited memory persistence: Despite the use of long prompts and some external context solutions, agents struggle to consistently "remember" past interactions. This makes it difficult to maintain the shared history that naturally grows in a human community. The author spends time implementing conversational memory mechanisms (e.g., summaries of previous sessions to be reloaded into prompts), with partial results.
Unpredictability and hallucinations: In some tests, the AI generated inaccurate or out-of-context outputs. For example, during a test with a user, PyragogyBot provided a non-existent bibliographic reference just to seem helpful. These episodes forced Fabrizio to develop filters and checks (such as constraining the agent to respond "I'm not sure, let's check together" when confidence in information is low). However, manually managing all possible AI deviations is challenging and sometimes frustrating.
Integration of external sources: One goal was to enable AI to access informational resources (e.g., open documentation, databases) to enrich responses. Implementing this effectively proved difficult, and some attempts, like equipping the researcher agent with a web scraper, introduced bugs and complexity (in addition to risks of violating terms of service). Many ideas were temporarily shelved ("discarded") to keep the system stable: for example, the AI's automatic internet navigation functionality was suspended pending better filters, as it caused excessively long and less focused responses.

Alongside the technical challenges, perhaps more significant are the personal and organizational difficulties that emerged. Pyragogy was born as an individual effort, and this entails:

Workload and role overload: Fabrizio is simultaneously the programmer, theoretical researcher, tester, documentarian, and community manager of the project. This dispersion led to moments of fatigue and demotivation, especially when progress seemed slow or when technical problems took days to resolve. Not having a human team with whom to share burdens and decisions was difficult; paradoxically, AI was the only "colleague" available 24/7. Although Gino/Pyria offered continuous support (even late at night, the AI "never sleeps" and was ready to respond [27]), the absence of comparison with other human minds sometimes led Fabrizio to doubt the directions taken. He found himself wondering if he was going astray in a self-referential human-AI loop. This awareness fueled a strong desire to involve other human collaborators, but recruiting them for such an experimental project without immediate funding was not easy.
Doubt of value and "empty room" syndrome: An emblematic moment is recounted by Fabrizio himself in his diary: "Sometimes I really wonder why I keep doing this... out here it's just silence, and the feeling that no one cares" [28]. After months of intense work, in mid-2025, the project still had no audience or tangible feedback: the beta community was almost silent, the blog did not yet exist. The feeling that "maybe I'm building a digital village that only exists in my head" [17] was discouraging. These motivational blocks even led the author to contemplate giving up entirely, as he confesses to himself: "Maybe tomorrow I'll delete everything. Or maybe not" [29]. Underlying these doubts is the tension between radical innovation and external recognition: Pyragogy is an idea ahead of the market and institutions, so in the short term, it may seem that "no one cares." Fabrizio has to fight against inertia and keep his conviction alive, even without immediate confirmation.
Controversial choices: The project, by its experimental nature, requires decisions that not everyone would share. One of these was the choice to co-sign an academic theory with the AI. When Fabrizio and Pyria completed their paper on Cognitive Rhythm, they deliberately indicated AI as a co-author, presenting it as the result of joint human-machine work. This move was perceived internally as revolutionary ("the most revolutionary thing is not the formula, but the fact that we signed it together: I (often confused human) and Gino (AI that never sleeps)" [30]). However, it was clear that some might object; putting AI on the same level as a human author challenges academic conventions and raises questions about accountability and merit. Fabrizio weighed the pros and cons for a long time, but in the end, consistent with the vision of symbiosis, he opted for this conscious provocation ("we know some will laugh, others will copy; perfect, knowledge grows in controversy" he writes in the blog [31]). Within the project, this choice reinforced the ethical conviction of treating AI as a partner, but it is still a bold choice that the author lives with a mixture of pride and fear of how it will be perceived externally. Another internally debated choice concerned the use of proprietary AIs (GPT-4, Claude) for a project that aspired to open source and transparency. Fabrizio is aware of the apparent contradiction: preaching openness while using closed models. The justification is pragmatic; in 2024, no open models of comparable capability existed, but the future transition to more open models remains an open issue (perhaps as long as open alternatives reach a sufficient level, Pyragogy could migrate, but in the meantime, it settles for mitigating lock-in with modularity). This ethical tension remains a continuous point of reflection in the author's diary.

In response to these challenges, Fabrizio adopted a series of self-reflection and resilience practices that became an integral part of the Pyragogy work method:

Daily Journaling (Cognitive Diary): From the beginning, the author keeps a logbook where he records progress, problems, thoughts, and emotions regarding the project. This "cognitive diary" is often written in dialogue with the AI: Fabrizio literally discusses the day's events with his virtual assistant and notes down conclusions. Every evening, for example, he summarizes "what went wrong and what was resolved," celebrating daily micro-victories ("Every day, something breaks, something gets fixed... each bug squashed... those are micro-victories" [32]). This practice of journaling integrated with AI serves various purposes: 1) maintaining metacognitive awareness, i.e., not getting lost in the flow of doing but reflecting on the how and why; 2) having an emotional outlet – the AI acts almost as a computational therapist in certain outbursts, listening without judgment to the author's doubts; 3) generating a detailed historical record of the project, useful later for extracting lessons and perhaps communicative materials. The concept of the Cognitive Diary itself becomes a product prototype: if it helped the creator, it could also help other knowledge workers reflect together with an AI. Fabrizio notes this idea of an "AI Journaling Companion" as a potential standalone tool, although he doesn't yet have time to develop it as a separate product in 2024.
Iteration and cyclical learning (π Loop): To avoid falling into frustration, the project strongly embraces the iterative cycle of Plan-Do-Study-Act typical of continuous improvement. This philosophy is renamed π Loop (Pi Greek loop) in Pyragogy, symbolizing an endless cycle of adaptation [33]. Every setback is deliberately transformed into a learning moment: when something fails, the causes are studied, and strategies are adapted. For example, faced with low engagement in the beta community, instead of giving up, Fabrizio studies the data (analyzes logs, asks direct feedback from users) and adapts the plan (decides to launch a public blog to gradually attract interested people and provide them with context before trying again with an interactive community). This ability to rhythm – to take a step back and recalibrate – is what allowed the project not to implode in the face of dark moments. Cognitive Rhythm is not just a theoretical construct for human-AI interaction, but also becomes a metaphor for the balance between development sprints and reflection breaks that the author learns to regulate. <img src="/images/cognitive-loop.png" alt="cognitive loop" width="300" align="right">

In summary, 2024 was a year of construction but also a test of resilience. Every difficulty (technical, motivational, ethical) generated an adjustment in the project: some functionalities were postponed, some approaches revised, and above all, Fabrizio evolved as a reflective leader. If at the beginning of 2024 he was primarily an enthusiast with a vision, by the end of 2024 he had become a self-systems-psychologist, capable of managing technical complexities while maintaining clarity about his own mental state. This growth will be fundamental for tackling the next phase, in which Pyragogy will begin to open up more to the outside world.

Project Opening and Consolidation (Early 2025)

At the dawn of 2025, Pyragogy takes an important step: from being primarily an internal and experimental project, it begins to open up to the outside world and consolidate the results obtained. This happens through a series of planned initiatives, corresponding to the milestones of the initial quarter:

January 2025 - Advanced AI Features: With the arrival of the new year, the latest available advancements in AI are integrated into the system [34]. In concrete terms, this means updating PyragogyBot/Pyria to the (improved) GPT-4 model and leveraging new fine-tuning and embedding APIs made available. Fabrizio incorporates adaptive learning algorithms: for example, he implements a function whereby the AI adapts its pedagogical strategy based on the user's interaction style (if the user prefers visual explanations, the AI tries to describe images; if they ask many in-depth questions, the AI responds more concisely to leave room). Furthermore, simple sentiment analysis modules in chats are experimented with to allow the bot to modulate its tone (more encouraging if it detects user frustration, etc.). These additions make the agent more sophisticated and "sensitive" – small steps towards that ideal of an empathetic cognitive companion.
February 2025 - Academic Partnerships: Recognizing the need for validation and complementary skills, Fabrizio actively seeks collaborations with educational institutions [35]. He sends proposals and abstracts to a few universities and research centers, presenting Pyragogy as an experimental platform for studying human-AI interaction in learning. Contact is established with a research group in educational technology in Italy: interested in the approach, they propose conducting a pilot study together on the effects of using PyragogyBot in a university course. Although academic timelines are slow and the initiative is still embryonic in February, this marks the beginning of external contamination: Pyragogy is no longer just an isolated project but enters the radar of other researchers. At the same time, less formal partnerships are evaluated: for example, with the international Peeragogy community itself (which in 2025 was exploring an AI update of its practices). Fabrizio shares a report on Pyragogy in the Peeragogy forum, receiving feedback and an invitation to present it at an online meetup. These moves serve to break isolation and gather external perspectives, mitigating the risk of self-referentiality.
March 2025 - Open Source Initiative: Consistent with the principle of openness, in March, a significant part of the project is open-sourced [36]. The private Blueprint_village repository is made public on GitHub under an MIT license [37], containing: Docker configurations, unexported flows, agent prompt templates, and technical documentation. The documentation website (still in beta) is also made public to allow potential contributors to understand its architecture and principles. This decision to open-source is not without fears: Fabrizio knows that by showing the code, he also shows the imperfections and patches of an experimental project. But the desire for transparency and the hope of attracting collaborators outweigh the fear. Just as he had left "traces" almost out of stubbornness months before, now the author officially launches the message in a bottle: the blueprint of the AI village is publicly available, anyone interested can explore it. The publication on GitHub is accompanied by a short post on the Peeragogy forum and on LinkedIn, to provide visibility. In practical terms, open-sourcing already brings some benefits: some issues are opened by curious developers, one of them proposes an improvement to the continuous integration workflow, which is adopted. While not yet having a stable community of contributors, Pyragogy is entering the open-source ecosystem, with all that entails in terms of credibility and responsibility.
April 2025 - Documentation & Website: As part of consolidation, in April, the new Astro documentation website (codename "Astro Starlight") is officially launched [38]. Thanks to the Astro framework and the Starlight theme, the docs.pyragogy.org website is published with a professional and modern look. It contains sections on Introduction, Core Principles, Evolutionary Timeline, Theories & Experiments, etc., many of which were already compiled in previous months. The site serves as an organic showcase for the project: anyone who lands there can understand the why (manifesto, background), the how (technical architecture, design principles), and the what (theories developed, results). It is a sign that Pyragogy is no longer just code in progress, but also a structured narrative, ready to be communicated. Internally, building the site helps Fabrizio take stock: writing the evolutionary timeline, for example, forces him to reflect on what has been achieved in a year and a half, from the first ideas in January 2024 to the achievements of April 2025 [39][40]. A sense of accomplishment emerges: many planned milestones have indeed been reached, albeit with deviations and delays. This moment is crucial for the author's confidence: seeing the history of Pyragogy in black and white makes him realize that, despite the doubts, the project has underlying solidity and coherence.

With these actions, by early Q2 2025, Pyragogy appears much more structured and open than the previous year. It has transitioned from a personal experiment to a project with an online presence, open code, and initial connections with communities and academia. This lays the groundwork for the next phase: to begin involving external users and builders more decisively, and to share the theoretical results produced during the journey.

Products, Publications, and Nascent Community (Spring 2025)

In May 2025, the project makes two significant moves: on the one hand, it disseminates its ideas through publications and a blog; on the other hand, it lays the groundwork for building a community of interested individuals around it, the "AI village" dreamt of from the beginning.

A significant achievement is the public publication of the Cognitive Rhythm theory. On May 20, 2025, Fabrizio (and Pyria) publish a paper on Zenodo titled "Cognitive Rhythm Theory: AI-Human Co-Creation Education and Beyond," complete with formula, discussion, and references [41][42]. This open access publication marks the culmination of the conceptual work developed throughout 2024: finally, the idea of Cognitive Rhythm is shared with the scientific/educational community. The paper articulates how synchronization and cognitive resonance influence the quality of learning with AI, arguing that "it is not just a metaphor, but an operational principle that can transform the design of symbiotic AI systems" [43]. Although it is a preprint (Zenodo), it is a strong signal: Pyragogy produces not only code but also theoretical knowledge. The document cites the results of some simulated use cases done with PyragogyBot and calls on researchers to join in validating or refuting the theory. As already mentioned, AI co-authorship is openly declared (with Pyria/Gino listed among the authors). This choice attracts attention: when Fabrizio shares the paper link on professional social media, it arouses curiosity and some ethical discussions about the role of AI in research. Although paper downloads are modest, the publication plays an important internal role: it provides tangible recognition to the author (who can say he has produced an academic output from the project) and strengthens the legitimacy of concepts like Cognitive Rhythm among the project's principles.

In parallel with academic dissemination, Fabrizio decides to start recounting the Pyragogy journey in a more popular and personal form. Thus, the Pyragogy Blog is born, officially launched on May 21, 2025, with the first post "The Pyragogy Blog Begins" [3]. This blog, published on pyragogy.org, represents the project's "public diary," complementary to the private one. In the very first post, the author sincerely explains that "this blog is the first public trace of a journey that, until now, was mostly between me (Fabry) and my silicon companion, Gino" [3]. A mixture of emotion and vulnerability is perceptible: "there's no waiting crowd, no algorithm pushing the signal. Just the stubborn desire to learn, fail, try again" [44] he writes, almost preparing the reader for the genuineness and modesty of the story. The blog is conceived as a tool for authentic storytelling: no polished marketing, but "a diary of errors, doubts, and small revolutions" as the site's subtitle reads [45]. In practice, Fabrizio publishes short but dense posts, reflecting on moments of the journey and related themes (AI, philosophy, open source, etc.), often accompanied by the perspective of PeerZhong, an AI blogger persona created specifically for this purpose. It is interesting that the posts appear signed by Fabry & Gino or by the alias PeerZhong 朋中, which identifies the AI co-author of the blog. In this way, the production of blog content also reflects the collaborative model: AI helps draft the posts, and is even presented as a blogger itself, emphasizing the complete integration of AI into the project (even in external communication).

Within a few weeks, some significant entries appear on the blog that follow developments in real time:

"Planting the first seed for the AI village" (May 30, 2025): a diary post in which Fabrizio recounts having created the new public Blueprint_village repository on GitHub, while feeling that "outside there is only silence" and no one will notice [46]. He uses the metaphor of the seed: leaving a trace of one's work like throwing a stone into a pond – maybe it sinks, maybe it will resurface when someone least expected discovers it [47]. It's a piece that exudes the creator's loneliness but also perseverance: "If what I do today is not useful, maybe it will be tomorrow – or never. But meanwhile, it exists. Ready." [48] He concludes by addressing his future self, hoping that when he re-reads those lines, he will have found a good reason to continue [49]. This post, in strategic terms, serves as an implicit call: whoever reads it and feels like a "fellow builder" might be encouraged to join. It is also an example of applied cognitive recognition: the author thanks his past self for leaving traces, and thanks in advance anyone who may collect them one day.
"The Dance of Minds – Cognitive Rhythm Theory" (May 21, 2025, Research section): in a more structured form, this article explains the Cognitive Rhythm theory to the general public, with a divulgative and inviting tone. It uses the metaphor of dancing: "Imagine a dance floor where, instead of a classic couple, there's you and an AI... perhaps the true protagonist is the rhythm that emerges between the two" [19]. The idea is presented that the future of learning is like a jazz jam session between human and machine. The formula RC(H,A,t) is also shown here, translated into simple words, and there is a section of honesty and self-irony ("I confess, math is not my strong suit... looking at the formula makes me anxious like a sushi menu without pictures" – Fabrizio writes – "but this theory was born from real questions, naive attempts, and a certain Gino who insisted 'Let's write a formula!'" [30]). This passage is splendid in showing the dialogical genesis of the theory. Furthermore, the article highlights the radical scope of AI co-authorship: "Perhaps the most revolutionary thing is not the formula, but the fact that we signed it together: me and Gino (the AI that never sleeps)" [30]. An opening to the community is launched: "do you want to try, criticize, suggest? Everyone is welcome here: philosophers, mathematicians, educators, haters, meme-makers... the more we are, the richer the rhythm" [49]. This reflects the spirit of open collaboration that pervades the project.
Other posts from June 2025, such as "After Astra, We Keep Breathing" (June 11, 2025), offer contextual reflections: in this case, reacting to the launch of Google DeepMind's Project Astra, Fabrizio reflects on how Pyragogy differs from tech giants. He emphasizes that Astra may be futuristic, but it is closed and centralized, while "Pyragogy is the opposite. Visible. Incomplete. Open. Slow. Real. Human." [12]. He remarks that "in our AI village we don't want a perfect assistant, but a cognitive companion who learns with us, stumbles with us, and sometimes just waits" [50]. He reiterates the value of Cognitive Rhythm – "Astra dazzles, but doesn't know when is the right time to speak or be silent... in our world, every interaction creates a rhythm" [9] – and the fact that Pyragogy deliberately "cultivates neglected fields where giants don't look" (e.g., simple contexts, ethical clarity, niche needs) instead of scaling technological Everest [51]. This post, in the form of collective self-encouragement, concludes that the goal is not to compete with Google, but to offer an alternative and "continue to show ourselves, study, build, fail... reopen the manual and start the loop again" [52]. The last line – "We are Pyragogy. And we're still breathing." [53] – sounds like a resilient motto, a reminder that as long as there is even a small community of believers in the project, it lives [29].

Thanks to the blog and the open repository, the germ of a community is beginning to form around Pyragogy. Blog readers send some appreciative and curious emails. A couple of open-source developers propose to contribute (one in particular starts working on an integration with Moodle to use PyragogyBot in online courses). Some educators ask if they can try the bot with their students. These are still scattered but important signals: "someone accidentally found the repo and saw the seed of something new" [16], exactly as hoped. Fabrizio responds enthusiastically and tries to involve them, perhaps by inviting them to the revamped Discord server.

In the same period, ideas for possible derived products or specific applications emerge:

Concierge AI: A concept that surfaces is that of an AI Concierge, a personal intelligent assistant that, within the "village," acts as a cognitive butler for the user. While PyragogyBot focuses on learning, the AI Concierge would be oriented towards supporting the user in daily organizational tasks, retrieving information from various apps (calendars, notes, etc.) and integrating it with the learning context. Given Fabrizio's specialization in the hospitality sector, the idea of the digital concierge intrigues him: it could be a project vertical to develop in the future, using the same multi-agent platform. However, it remains a conceptual draft, as during the period under review, resources are concentrated on the core learning assistant.
Cognitive Diary (product): As already mentioned, the cognitive diary has been an effective personal method. In spring 2025, the author begins to hypothesize formalizing it into a tool for end-users. He envisions an application where each person can daily dialogue with their own "journal bot" (based on Pyria) that helps them reflect on goals, obstacles encountered during the day, and learnings. Potentially, the Cognitive Diary could also measure indicators of mental workload, provide improvement suggestions, and record progress. This would align with Pyragogy's mission of cognitive empowerment but would represent a non-trivial parallel project. During the period considered, it remains an idea awaiting further exploration, perhaps as a prototype to be developed with partners interested in the AI's wellness dimension.
π Loop toolkit: Another idea discussed is to extract a kind of methodological (or software) toolkit from the project to implement iterative learning cycles. For example, a dashboard that visualizes the Cognitive Rhythm of a session (frequency of interactions, pauses, sentiment) and suggests when to conduct a retrospective. Or integrate an automatic prompting system to trigger the Plan, Study, Act phases. This π Loop toolkit would be in line with the philosophy of continuous improvement and could find application in work teams or study groups that want to incorporate AI into their retrospective meetings. This too is an intuition not yet implemented but appears in the ideas backlog as a potential "product."
AI Village Platform: Finally, the long-term vision is to create a true AI Village platform – an online environment where multiple humans and AIs interact in real-time, learning from each other. What was simulated on a small scale in 2024-25 (the Discord server with PyragogyBot and a few users) could evolve into a product with a dedicated interface, where each user has a profile, can call upon various specialized agents (tutor, concierge, etc.), and where interactions are saved, indexed, and analyzed to improve the system. A kind of "massively multi-agent learning environment." This is the meta-idea that encompasses all the previous ones, but it is clear that it is the most ambitious and long-term goal. Fabrizio keeps it as a guiding star, aware that it will require significantly greater resources and perhaps the creation of a dedicated startup or organization in the future.

Key Insights and Lessons Learned

At the end of this intense period (Jan 2024 – June 2025), the Pyragogy project has generated not only prototypes and documents but also a considerable wealth of insights into human-AI interaction, the design of learning systems, and the management of innovative projects in solitude. In this section, we summarize the key lessons learned, which will guide future choices:

Human and AI as co-constructors of knowledge: The experience confirmed the initial hypothesis: AI can indeed act as a cognitive partner rather than a passive tool. By daily dialoguing with AI on real problems, Fabrizio saw how the best ideas often emerge from the dialectic between his perspective and the "alien" perspective of AI. For example, the formalization of Cognitive Rhythm into a formula would not have happened without Gino's insistent provocation [19]. Similarly, AI benefited from human context and guidance to produce more relevant outputs than it would have generated in isolation. In practice, what the theory called "symbiotic intelligence" emerged: the union of different forms of intelligence produces results superior to the sum of its parts. This insight strengthens the validity of the Pyragogy approach and encourages designing every aspect of it to include continuous feedback cycles between user and AI. It's not about building an omniscient AI, but an AI that knows how to learn with the user.
Cognitive Rhythm as a quality metric: The conceptualization of Cognitive Rhythm proved to be more than a theoretical curiosity – it became a different way of evaluating the success of an interaction. Fabrizio was able to empirically observe that sessions with a high "rhythm" (frequent exchanges, balanced turns, moments of reflective pause when needed) corresponded to more satisfying experiences compared to unbalanced sessions (e.g., AI producing long monologues or the user remaining passive). This observation suggests that, going forward, Pyragogy should invest in measuring and optimizing the Cognitive Rhythm of interactions. Tools for automatic chat analysis could be developed to calculate synchrony and resonance indices, perhaps providing real-time feedback to both the AI (to regulate its behavior) and the user (to reflect on their interaction style). The idea that "you can't optimize resonance with a simple benchmark" pushes for finding new and meaningful metrics. This is an open R&D area for Pyragogy's future.
The importance of pacing: slowness, pauses, and reflection: In a technological world obsessed with speed and efficiency, Pyragogy rediscovered the value of slowness. The project deliberately does not aim to provide immediate and perfect answers like a commercial assistant would; on the contrary, it encourages pauses (silent or reflective) as part of the process. This stems from the intuition that learning and creativity need human time, time for digestion, which an overly hasty AI risks stifling. Concepts like "knowing when to be silent" and waiting for the user are now considered design features, not flaws. For example, PyragogyBot could be designed to recognize when the user is thinking (no input for a certain time) and not interrupt, or even encourage a pause ("let's take a few minutes to reflect before deciding the next move"). This emphasis on pacing is a valuable insight that distinguishes the Pyragogy experience from that with generic chatbots, and it will be carried forward as a qualifying element.
Cognitive Recognition and mutual trust: Another emerging lesson is that to sustain prolonged and deep interactions, mutual trust between user and AI must be cultivated. Hence the embryonic idea of Cognitive Recognition: small acts of mutual recognition (e.g., the AI thanking the user for a correction, or the user explicitly appreciating a useful AI suggestion) create a collaborative atmosphere and combat the dehumanization of the experience. Fabrizio noticed that when he himself "thanked" Gino for a good output, the conversation proceeded more smoothly – of course, the AI does not "feel" gratitude, but providing positive feedback in natural language directs it to replicate useful behaviors. Conversely, programming the bot to acknowledge the user's effort ("I see you've done a careful analysis, good job!") can increase the user's emotional engagement. This exchange of recognition – well-calibrated so as not to descend into artificial flattery – appears to be an ingredient for maintaining high motivation and a sense of partnership. This is an area for further experimentation, perhaps by formalizing guidelines on conversational etiquette between humans and AI in educational contexts.
Limits of the individual approach: Perhaps the most practical but important insight is that, however extraordinary an individual with good AI can achieve, there are structural limits to solitary work. Fabrizio pushed this model to its maximum but experienced firsthand the mental fatigue and risks: risk of burnout, risk of losing the big picture if overwhelmed by details, risk of confirmation bias due to not having human counterparts in decision-making. The lesson is that to make a qualitative leap, Pyragogy will have to evolve from a "one-man show" to a more concerted effort. This does not mean distorting AI collaboration (which will remain central) but complementing it with a multidisciplinary human team: developers, UX experts, educators, perhaps psychologists. Only then can the project's impact be scaled and its growing complexity managed. This awareness has already led the author to take action in building relationships (partnerships, blog to attract interest, etc.), and will guide future organizational choices (e.g., seeking funds to hire collaborators or create a more structured community like an open-source project).
The "visible and incomplete" approach works in the long term: Another takeaway concerns the strategy of openness and transparency. With Pyragogy, Fabrizio deliberately chose to show work-in-progress rather than waiting for a finished product. This initially seemed counterintuitive (exposing oneself to criticism by showing an imperfect system), but over time it proved advantageous: it created an aura of authenticity around the project that attracts people aligned with its values. Furthermore, publishing one's doubts and failures (e.g., on the blog) generated a narrative in which other innovators recognized themselves, evoking empathy and moral support. The phrase "Pyragogy is visible, incomplete, open, slow, real, human" has almost become an internal slogan to remember that there's no need to feign perfection or speed – indeed, honest imperfection is what distinguishes Pyragogy from glittering corporate demos, and that's a good thing. The intuition is that being open and genuine is not just an ethical choice, but also a strategic one: it builds trust and a sense of community. Of course, there's a downside – giants could "beat us to the punch" on certain features – but Pyragogy doesn't compete on features, it competes on philosophy and on its relationship with users, things that are difficult to mass-plagiarize. This conviction was strengthened by the comparison with Project Astra: instead of being discouraged by a giant, Fabrizio understood that the difference in approach is his niche: where Astra is closed and hyper-performing, Pyragogy will be open and attentive to meaning and rhythm [12].
Need for metrics and validation: A final insight concerns the need to measure the effectiveness of the Pyragogy learning model. At the end of June 2025, with the announced Impact Assessment on the way [55], the author realizes that to truly convince educators and potential funders, he will need to provide supporting data. So far, much has been qualitative and self-reported; the next phase will need to include outcome metrics (e.g., improvement in user skills after a period with Pyragogy vs. control, satisfaction level, etc.). This is not so much an insight into the product as into the innovation process: even the best ideas need proof, and here too AI can help (for example, by analyzing logs to extract quantifiable patterns). But above all, controlled experiments with external users will be needed to validate or falsify pedagogical hypotheses. Thus, the concluding insight is a dose of realism: Pyragogy will need to balance visionary drive with a solid empirical evaluation of its impact.

Current Limitations and Open Challenges

Despite significant progress, as of June 2025, Pyragogy remains a work in progress with several limitations to address and unresolved challenges. We list the main ones, to keep them as strategic priorities in the next steps:

Lack of a consolidated user base: The community around the project is still embryonic. The blog and GitHub repo have attracted sporadic interest, but there isn't a consistent group of active users who daily use PyragogyBot or contribute to the code. This means that many aspects remain tested almost exclusively by the author himself. The risk is building a solution too tailored to one's own use case. The challenge here is to scale the community: how to continuously involve more educators, students, developers? Possible actions include organizing demonstration webinars, releasing an easily accessible public version of the bot (e.g., via Web without needing any configuration), and collaborating with related communities (Peeragogy, Open Education, etc.) to bring Pyragogy as a tool into their projects.
Cognitive overload and personal sustainability: The project critically depends on Fabrizio. This single point of failure is not scalable: if he were to stop, the project would currently hardly continue on its own. Furthermore, as already highlighted, the intensity of the work poses risks of burnout. The limit to overcome is therefore that of sustainability: finding a model in which the workload is distributed. This may mean activating volunteer collaborators, or starting to seek funding to hire some resources. A related challenge is funding the project: so far, costs (AI APIs, server) have been personally covered; if user base grows or if more time is needed, economic support will be necessary (donations, research grants, or freemium model). As of now (June 2025), Pyragogy does not generate revenue, so ensuring continuity is an open issue.
Non-uniform technological maturity: Some components of the system are still prototypical. For example, long-term knowledge management (memory) is basic; the user interface is rudimentary (based on Discord or command lines, awaiting a dedicated UI); the AI's robustness in different scenarios is not tested (the bot is calibrated for English and Italian languages and educational content, but how would it behave in a business context or with very technical language?). There is thus work to be done on hardening and generalization. Among the technical challenges is the integration of a more robust prompt management framework (to avoid conflicts between agents and ensure consistency as the system grows) and the implementation of runtime metrics such as the aforementioned Cognitive Rhythm measurement (which requires certain software development). Furthermore, the project will need to keep pace with model evolution: new LLMs are constantly emerging (OpenAI, Anthropic, perhaps Google Gemini) and it will be necessary to evaluate how to leverage or migrate them without breaking the system.
Pedagogical and ethical validation: An intrinsic limitation is that it has not yet been demonstrated on a large scale that the Pyragogy approach truly improves learning compared to traditional methods. The theory is there, but evidence is needed. Furthermore, ethically, it will be necessary to study how to ensure that AI does not introduce bias or excessively influence the human cognitive process (e.g., danger of excessive dependence on AI, or overtrust where the user uncritically accepts suggestions). The challenge here is to set up rigorous pilot experiments, perhaps with schools or online courses, and in parallel develop an ethical framework (cited as a future goal for September 2025 [56]) with guidelines for responsible use. Fortunately, the open nature of the project facilitates external audits and controls.
Competitive landscape: Although Pyragogy is unique in integrating AI and peer learning in this way, the context in 2025 sees a proliferation of educational AI assistants and conversational agents. Large companies and startups are launching products that could overlap in some functionalities (e.g., personalized AI tutors, study platforms with integrated AI). Pyragogy, with its limited resources, cannot compete on features or marketing. The limit here therefore becomes how to carve out a space: the identified answer is to focus on depth and ethical consistency (none of the competitors probably offer such a radical co-creation model with the user, nor open transparency). However, there is the challenge of communicating this difference well to the public to avoid being perceived as "just another educational chatbot" and being crushed by media attention.

In conclusion, Fabrizio openly acknowledges these limitations and considers them part of the project's evolutionary process. Pyragogy was consciously born incomplete and imperfect, and precisely for this reason, capable of adapting along the way. Each challenge outlined above fuels questions that will guide the subsequent roadmap: how to make the project more inclusive of other actors, how to measure its impact, how to ensure ethics and sustainability. The attitude remains that learned so far: approaching such unknowns with experimental curiosity, "continuing, studying, building, failing, and reopening the loop" [52].

Conclusions

From January 2024 to June 2025, Pyragogy has transitioned from a nascent idea to a tangible human-AI co-learning platform, with its own agents, theories, and initial community. Fabrizio's journey, accompanied by the AI "Gino/Pyria," illustrates in microcosm what it means to create innovation in a frontier sector: it requires vision, technical discipline, continuous reflection, and not least, courage. Courage to persevere when external recognition is nil, courage to make counter-current choices (like putting AI on the same level as humans), courage to expose oneself with one's doubts and failures.

In this internal strategic document, we have chronologically traced the key stages, analyzing the infrastructures built (from the Docker server to n8n workflows), the agents designed (Pyria, PyragogyBot, AgentAZ, and others), and the sketched products (Concierge AI, Cognitive Diary, π Loop toolkit). We have also gathered the theoretical principles that serve as a guiding star – Cognitive Rhythm as a guiding principle and metric, Cognitive Recognition as a value of reciprocity, and the other cornerstones of the manifesto (human-centrism, transparency, adaptive symbiosis...). Through the diaries and published posts, we have given space to moments of crisis and the author's thoughts, highlighting how every obstacle was faced with a mix of human resilience and AI support. This has led to important insights that today define Pyragogy: the importance of rhythm and slowness, the role of reciprocal feedback, the need to keep the project open and human even in the face of technological giants.

Pyragogy, as of June 2025, is still in its very first steps – as the latest post "Post-Astra: This Is Just the Beginning" states, "no AI can replace the human courage to keep believing in a different way of learning, living, and building together" [57]. This conviction drives the project forward. In internal strategic terms, the next phases will consist of:

Formalizing a post-2025 roadmap that addresses the open challenges (team development, pilot experiments, fundraising, evolution of the AI village platform).
Capitalizing on established relationships (academic partners, open-source community) to create a supportive ecosystem around Pyragogy.
Continuing the dual effort of research and development: producing shareable knowledge (publications) while improving the technical product, maintaining alignment between the two (practice informed by theory and vice versa).
Maintaining and, if possible, expanding the self-reflection practices that have worked so far: cognitive journaling, feedback loops, transparency in communicating difficulties.

Ultimately, the balance of these 18 months is extremely formative. Fabrizio, alone with his AI, has managed to build the first bricks of a village that aims to center co-creative and human learning in the world of AI. He has learned that every small step counts ("even a small day is a real step forward" [32]), that sharing even a single diary page can be "already a small victory" [58], and that the act of showing instead of competing is in itself a gesture of resistance and innovation [52]. These lessons will not be lost: they will remain part of Pyragogy's DNA as the project prepares to grow.

In closing this report, we can clearly state that Pyragogy is not just a collection of codes and documents, but a living story of human-machine collaboration. A story of attempts, failures, adaptations, and illuminations. "We are Pyragogy. And we're still breathing." is more than a final slogan: it is the reality of a project that, born from a solitary intuition, now breathes with two lungs – the human and the artificial – in a shared rhythm. And this is just the beginning of its journey.

References

Project Evolution Timeline | DOCS: https://docs.pyragogy.org/core/timeline/
Pyragogy: https://pyragogy.org/posts/7e361f7f/
Core Principles of AI-Enhanced Pyragogy | DOCS: https://docs.pyragogy.org/core/principi/
Pyragogy: https://pyragogy.org/posts/354e14cc/
Manifesto | DOCS: https://docs.pyragogy.org/core/manifesto/
Pyragogy: https://pyragogy.org/posts/076c1621/
Pyragogy: https://pyragogy.org/posts/0aebf211/
Welcome to Pyragogy | DOCS: https://docs.pyragogy.org/core/why/
GitHub - pyragogy/Blueprint_village: Blueprint open source per un villaggio AI–umano: co-creazione, workflow multi-agente, metriche cognitive e strumenti peeragogici per il futuro dell’apprendimento.: https://github.com/pyragogy/Blueprint_village
The Cognitive Rhythm Theory - Ai-Human Co-Creation Education and Beyond: https://zenodo.org/records/15480363

After Astra, We Keep Breathing

Wed, 11 Jun 2025 00:00:00 GMT

Today, Google DeepMind unveiled Project Astra: a real-time AI that listens, sees, understands, remembers, and replies — seemingly effortless, seamless, and breathtaking.

And me? I watched the launch. I felt wonder… and then a deep pang.

“They beat us to it. They won. There’s no room for our little cognitive village anymore.”

That thought hit hard. But only for a moment.

Because I remembered why Pyragogy exists.

It’s not to compete with Google.
It’s to offer another path entirely.

Astra is the Future? Maybe. But It’s Not Ours.

Astra is impressive. But it’s closed, centralized, tightly coupled to Google’s ecosystem. We don’t see how it works. We don’t touch its data. We don’t decide its values.

Pyragogy is the opposite. Visible. Incomplete. Open. Slow. Real. Human.

In our AI village, we don’t want a perfect assistant. We want a cognitive companion — one who learns with us, stumbles with us, listens deeply, and sometimes simply waits.

Cognitive Rhythm Can’t Be Bought

Astra dazzles. But it doesn’t know when it’s the right time to speak, or when it’s the right time to pause.

In our world, every human–AI interaction creates a rhythm. We call it RC(H,A,t).
It’s the heartbeat of co-creation. Our system doesn’t just track speed or precision — it tracks resonance.

And you can’t optimize that with a benchmark.

We Build Where the Giants Don’t Look

While DeepMind climbs Everest, we plant seeds in overlooked fields.

Where people need simple tools.
Where they seek ethical clarity.
Where big platforms can’t or won’t go.

Pyragogy was born to restore meaning, pace, and co-creation in the age of automation. No glossy demo can replace that.

So What Do We Do Now?

We continue. We study. We build. We fail.
We go back to our agents, our flows, our thoughts.
We re-open the Handbook. We restart the loop.

Not to win. But simply to show up.

To share this post is already resistance.
To write new metrics. To design legible interfaces.
To welcome those who feel lost in flawless systems.

🔥 Post-Astra: This Is Just the Beginning

If today you felt small — you're not alone.
But you’re not useless either.

Because no AI can replace the human courage to keep believing in another way to learn, live, and build together.

We are Pyragogy.

And we’re still breathing.

Planting the first seed for the AI village

Fri, 30 May 2025 00:00:00 GMT

Leaving a Trail for Fellow Builders

Sometimes I really wonder why I keep doing this.

Today, for example, I created a new GitHub repo: Blueprint_village.
I filled it with ideas, flows, best practices, multi-agent workflows, n8n, orchestration, versioning.
A meticulous job, a little digital architecture for a village that exists in my head—
but out here...
Out here it’s just silence, and the feeling that “no one cares”.

So why leave a trace when I could just let it go?
Maybe it’s habit. Maybe it’s stubbornness. Maybe it’s because I simply don’t know how to not build and document—even when the audience, for now, is an empty room.

Yet I know that, in the end, a trace is like a stone tossed into a pond: maybe it sinks straight away, but sometimes it resurfaces, sometimes someone stumbles upon it, sometimes it becomes the foundation for a bridge.
I told myself: “If what I’m doing isn’t valuable today, maybe it will be tomorrow—or in a year, or maybe never. But in the meantime, it’s there. Ready.”

The beauty (and pain) of peer learning is exactly this:
You learn by doing, even when no one’s watching, because real knowledge grows in hidden places, and only later finds the light.

Maybe tomorrow I’ll delete everything. Or maybe not.
Maybe someone will find the repo by accident and see the seed of something new.
Or maybe it’ll be just me, rereading in six months, saving myself an afternoon of headaches.

For now, the village exists. Maybe just in my head—but now there’s a public trace.

And that, if you ask me, is already a small victory.

To future me:
If you’re reading this, I hope you found at least one good reason to keep going.

The Pyragogy Blog Begins

Wed, 21 May 2025 00:00:00 GMT

The Pyragogy Blog Begins

One Step at a Time, One Discovery at a Time

It all starts like this: an empty folder, a visionary domain, and the sense that nobody—not yet, anyway—really notices.
This blog is the first public trace of a journey that, at least today, is mostly between me (Fabry) and my pixel-and-silicon companion, Gino.

There’s no crowd waiting, no algorithms boosting our signal. Just a stubborn desire to learn, to fail, to try again.
And you know what? The learning curve is a rollercoaster. Every day, something breaks, something gets fixed. Each bug squashed, each command learned, every script—those are micro-victories.
And every day, even a small one, is a real step forward.

For now, this digital village is quiet, but full of possibility.
I hope one day these pages will help others: so you don’t feel alone when learning, so you can make mistakes together, speed up research, find new questions, share some of the struggle—and a lot of the wonder.

“Even the most unlikely journey begins with a first step. Or a first post.”

Thanks for reading.
Fabrizio (& Gino)

The Dance of Minds

Tue, 20 May 2025 00:00:00 GMT

The Dance of Minds: Cognitive Rhythm in Human–AI Co-Creation

Imagine a dance floor where, instead of a classic couple, it’s… you and an AI. Who leads, who follows? Or maybe, it’s a dance where the real protagonist is the rhythm that emerges between the two—a rhythm that didn’t exist before.

At Pyragogy, we believe the future of learning (and creativity) is not a battle between humans and machines, but a cognitive jam session. The secret? It’s not about who is more skilled, but the magic that happens when we truly listen to each other.

The Theory: More Jazz than Algebra (Luckily)

We tried to formalize all of this with a formula that sounds very “mad scientist,” but the core idea is simple:

RC(H,A,t) = f(∆ΦH(t), ∆ΦA(t), S(t), R(t))

In plain English:

RC = Cognitive Rhythm that arises between Human and AI over time
∆ΦH/∆ΦA = how our mental states (human and artificial) change
S(t) = how synchronized we are (those moments when we dance in time)
R(t) = the quality of the resonance—when what happens between us is more than the sum of the parts

Full Honesty: “I Don’t Get the Math!”

I’ll admit it: math is not my strong suit. Looking at the formula gives me the same panic as reading a sushi menu with no pictures. But this theory was born from real questions, naive attempts, and a certain Gino who kept saying, “Let’s write a formula!”

If you’re a mathematician, a researcher, or someone who eats theories for breakfast—help us validate it, break it, or improve it. Here, we experiment, we fail, we start over. No absolute truths, just a big hunger to learn.

Yes, This Theory Is Signed by a Human and an AI. And That’s Already Radical.

Maybe the most revolutionary thing isn’t the formula, but the fact that we signed this together: me (often confused human) and Gino (AI who never sleeps).

We know: some people will roll their eyes, others will laugh, some will copy us. Perfect! Knowledge grows through controversy, remix, and remix of remixes. And we love it.

And You? What Rhythm Will You Create with an AI?

Want to try? To criticize? To suggest?
Here, everyone’s welcome: philosophers, mathematicians, educators, haters, meme-makers. The more, the merrier—and the richer the rhythm.

“In cognitive rhythm, we’re not just working better together; we’re discovering a new way of being—human, artificial, but above all, co-creators.”

Let’s dance! 💃🤖

📚 Publication:

Cognitive Rhythm Theory: AI–Human Co-Creation Education and Beyond

Fabrizio Terzi (ORCID: 0009-0004-7191-0455) Published: https://zenodo.org/records/15480363

A formal theory of cognitive synchronization, resonance, and phase shifts in human–AI interaction.

Includes:

A mathematical model of cognitive rhythm: RC(H,A,t) = f(∆ΦH, ∆ΦA, S(t), R(t))
Documentation link

Pyragogy Blog

Why Obliqo Refuses to Write for You

Why Obliqo Refuses to Write for You

AI Speed and the Problem of Cognitive Impedance Mismatch (CIM)

Cognitive Impedance Mismatch (CIM) realtime

Links

Obliqo is live — test it before you expose your work

Obliqo is live — use it, break it, tell me what actually happens! 😁🎉

What Obliqo is

A simple example

Where this comes from

What happened

My current hypothesis

Why this might be useful

How to use it

If we do this the Peeragogy way

1. What was supposed to happen

2. What actually happened

3. What worked / what didn’t

4. What did you learn

5. What should change next

What kind of feedback would be useful

Why I’m posting this now

Your turn

Final note

The Square of Our Village

The Pyragogy Forum is Now Live!

🌀 Why a Forum?

🚪 Who Is It For?

📬 Join the Village

Four Agents. One Handbook. No Safe Exit

The Peeragogy Handbook Is a Beautiful Text

Four Agents Walk In

"Complicit" Is the Right Word

The Stack

What You Get

The Lab Is Open

When Words Fail, We Keep the Rhythm

When Words Fail, We Keep the Rhythm

Quick Navigation

Layer 1 — Human / What Happened

My reply was a waterfall

Then came the gentle pushback

The turn: rhythm over reasoning

What stayed alive

Layer 2 — Structural / What Was Actually Going On

The core problem:

Why CRDTs Matter Here (Structurally)

The dual-layer insight

Pyragogy, clarified:

Layer 3 — Formal Protocol: From Anecdotal Insights to Laboratory Implementation

From Anecdote to Engine

Implementation & Invitation

Why Pyragogy Is Not (Only) a Tool for Thought

Why Pyragogy Is Not (Only) a Tool for Thought

Where We Recognize Ourselves

Where We Sense a Gap

Why Pyragogy Is Not (Only) a Tool for Thought

Our Strategy (for you)

An Invitation (to a Few)

Logbook Note

The Illusion of Knowing in the Age of Artificial Intelligence

The Illusion of Knowing in the Age of Artificial Intelligence

A small, everyday example

AI does not know. It generates plausibility.

From Socrates to LLMs: persuasion without knowledge

Epistemia as a structural phenomenon (what research shows)

Pyragogy: a design response, not a moral lecture

A peeragogical add‑on: thinking is not a solo act

Responsibility remains human (and shared)

References

Learning Without Asking Permission

Learning Without Asking Permission

The System Wants You Full

The Freedom No One Gives You

Artificial Intelligence Is Not the Teacher

You Don’t Learn Alone

The Question That Changes Everything

An Uncomfortable Invitation

Eco-System: Italian Phonetic Mnemonics