A working paper  /  FileMap × MONK

An AI partner for how people actually learn.

A space you can see all at once, and an agent that reaches out first. Built to help design education scale without losing the human.

Yildirim Yazganarikan (Leo) Founder, FileMap · creator of MONK filemap.com

The premise

As a company we are both excited and concerned about the pace of AI. We foresee a major impact on the workforce, and we believe education will matter more, not less, in the years just ahead.

We think conventional education has to be supported by AI, and that structuring agentic AI and human-AI dynamics is the strategic question of this moment. Without that integration done carefully, we expect the next generation to fall behind very quickly. So we built two things that work as one: a visual workspace, and a proactive agent that lives inside it.


Two instruments

The space

FileMap

a zoomable, infinite visual workspace

It turns the hierarchy of folders into a single plane you take in at a glance, the way a design studio reads its wall of pinned-up work. Every file, sketch, model, and reference is an object you place, resize, and relate in space.

  • Uses visual memory and spatial awareness to cut cognitive load
  • Shows the whole project, and the whole thinking process, not just the final picks
  • Already a shared workspace teams choose to gather around
The agent

MONK

a proactive, curious mentor persona

Where Claude, Copilot, or Notion AI wait to be asked, MONK reaches out first. It gets to know a learner over time, notices patterns, and picks its moment. It is designed to be a mentor, not a lookup box.

  • Proactive: it starts the conversation and builds habits over time
  • Intelligent, curious, analytical, insightful, by design
  • Asks questions, challenges thinking, encourages growth

Why a visual space

The studio wall, made infinite.

FileMap came out of a habit from architecture and design studios: pin work to a wall or lay it on a table, step back, and read the big picture at once. That single move, seeing everything together, is where connections happen. FileMap moves it into the file system so it never has to be rebuilt.

What is seen reminds and guides the person, both consciously and subconsciously.

There is real cognitive science under this. Carl Jung wrote that images and symbols form meaningful connections in the mind beyond conscious awareness, so a person laying out their own visual map can find a clearer path through their interests and goals. Daniel Kahneman's work on priming showed that visual cues steer our thinking and our decisions even when we never notice them. FileMap is built on that: the layout is a thinking tool, a narrative of what a person is chasing, that keeps nudging them toward it.


The proactive agent

Most AI waits. MONK begins.

Today's assistants are passive: they sit until a user knows what to ask. But a learner rarely knows the right question, or that a tool even exists. So MONK is proactive. It develops habits with a person over time, and it opens conversations rather than only answering them.

That changes what an agent can do in a classroom. A teacher's MONK can ask, week to week, about their teaching goals, help decide which to bring to the class, and choose how to deliver them, a question here, a resource there. Those goals then reach each student through their own personal MONK, in their own context.

feedback: goals, progress, motivation Teacher defines goals shares teaching ideals MONK proactive mentor as questions & resources Students each with a MONK FileMap - the shared visual workspace everyone acts on
The proactive cascade: a teacher's intent reaches students through their own agents, and comes back as signal.

What it is for

The goals we are actually building toward.

01

Personal, scalable support. Real-time help with learning, design, and picking up new tools, for the individual and the group, without trading away quality as it scales.

02

Tuned to the person. Actions and guidance shaped to each student's curiosity and motivation, using group dynamics and psychological principles.

03

Closer human ties. Help teachers state their goals clearly and deliver them well, and bring teacher-to-student and student-to-student communication closer.

04

A living library. With consent, build an academic and professional record of design tools and processes, so the whole class learns from what works.

05

Models that fit the context. Customize at the level of the person, the class, the department, and the school, guided by educator and administrator feedback.

06

A universal format. A model any institution can adopt, flexible across cultures and curricula, that still delivers concrete, measurable value.


In the studio

Two field scenarios.

How the space and the agent feel from inside a real design course. These are the interaction studies we work from.

Field note 01  ·  digital media design

Beliz turns a sketch into 3D, and the class follows.

Beliz is preparing a project for a digital media course. She drops a concept image onto her workspace, and her agent notices.

MONK

Hello Beliz. I see you added a new concept image. Is this your design?

Beliz

Yes, I drew it myself. I want to turn it into a 3D model, but I am not sure how.

MONK

I cannot build it for you, but I can point you to the right tool. You have used Rhino before. Do you want a detailed model, or a quick 3D preview?

Beliz

Quick. I just want to see it in 3D, I can add detail later.

MONK

Then try Krea's Image to 3D. It converts your drawing in seconds. Low detail, but ideal for a fast preview.

Beliz

That was really fast. It looks just how I wanted, simple and clear.

MONK

Does the folder's size and position reflect how important this is? The panel showing the model looks small for such a key element.

Beliz

You are right. (resizes it) Now I can see all the detail. Much better.

MONK

Since this is the heart of your project, maybe record the process as a short narrated video. You will document it and inspire your classmates. Shall I start the screen recorder?

MONK

Your video in the class channel is getting attention. Merve and Ahmet watched it and already want to try the same tools. Professor Hakan clicked the link and praised your proactive approach.

One student's discovery becomes the whole class's, with consent, inside the shared space. Learning compounds.

Field note 02  ·  a full semester

Maya and Atlas organize a year, then re-organize it.

Maya, a third-year architecture student, opens the workspace for the first time. Her assistant, Atlas, appears.

Atlas

I see you imported your third-year course folders. Want me to help organize them?

Maya

Yes please. I already feel overwhelmed, there is so much to track.

Atlas

Let's clarify priorities. Which course will need the most attention?

Maya

The design studio, definitely. A year-long community center project.

Atlas

How do you usually work, through research, sketching, references?

Maya

Research and references first, then sketch. But I lose track of references and cannot find my early sketches when I need them.

Atlas

Then I will put the studio at the center as the main panel, with research, references, sketches, and models around it, and a shared section for sustainability that both your studio and sustainability class can reach. Everything stays visible even when you zoom in.

Maya

That sounds perfect. I love how you are connecting themes across courses.

Atlas

(two weeks later) You have added a lot on outdoor gathering spaces. Has your focus shifted?

Maya

Yes. After a site visit I realized the center should emphasize the outdoors. Even my professor said my project shows an unusually strong link between the technical side and the experience, I think it is because I can actually see those connections in my workspace.

The workspace grows and changes with the thinking. It is not storage, it is a place to think.


How it works

A feedback loop, with consent at the center.

Operation

A continuous loop

The system records human-to-human and human-to-AI dialogue and action, individually and in shared workspaces. Models and knowledge sources are developed and tuned for each workspace and user over time.

Evaluation

Each person's own goals

It works from each person's self-defined goals, teacher and student alike, tracking strengths, weaknesses, and learning desires against progress, and measuring satisfaction, learning outcomes, time saved, and quality, in real time rather than only end-of-term surveys.

Privacy

A vault and a boundary

  • Every user has a personal data vault and their own agent
  • The workspace-level agent stays fully separate
  • Consent is explicit, and protocols refresh it regularly

Field record

Already in real classrooms.

The visual workspace is not a proposal. It has been alpha-adopted in design studios, chosen by faculty because seeing a whole project at once changes how students work.

University of Maryland
School of Architecture, in studio under Assoc. Prof. Dr. Hooman Koliji, used for student project reviews.
ITU, Faculty of Architecture
Spring 2024-2025 studio, run with a class of 20 students.
METU · TOBB · YTU
Studio courses and workshops across three Turkish universities.

A universal format, abstract enough to fit any culture, concrete enough to deliver real value.


Open questions

The questions we are holding while we build.

These are drawn from our own internal discussions. We would rather show the real ones than pretend they are settled.

01

How can agentic workflow systems be integrated into education, honestly and well?

02

How can knowledge and experience be recorded, preserved, and shared efficiently through AI?

03

How can students, and teachers, be connected to new information and methods more strategically?

04

How can AI reveal the potential of students and teachers through well-timed questions and guidance?

05

How should the system sit within student-teacher, student-student, and teacher-administration dynamics?

06

Can learning styles and performance transfer across contexts, and how should consent for that be designed?

07

How do we reach these ideals within real limits: context windows, hallucination, fine-tuning, data strategy?

08

How should the architecture stay flexible enough for constant testing and iteration?


The mission

We are deeply interested in AI and education, and we want to take responsibility here. The goal is to begin a journey that questions the essence of learning itself, and to build the space and the agent that make it human.

Yildirim Yazganarikan (Leo)
Founder and CPO, FileMap · creator of MONK
filemap.com · yy@filemap.info
FileMap × MONK · a working paper on visual, proactive AI for design education