Hi, I'm Jesse

When I'm not spending time with my family, I'm usually doing one of two things, often both: living a little in the future, and trying to understand what this being-human thing is actually about.

That usually means thinking causally about why people behave the way they do inside the worlds they're in.

I build and do things to test whether my ideas about these systems are actually true.

I'm self-taught and self-directed. When I get obsessed, I go learn and I do the thing.

Recently, this looks like iteratively developing my daily commitment system (Caring Ledger) using Claude Code.

If you want the concrete version, scroll down. I show what I'm working on and why.

I'm deeply preoccupied with individual harm reduction and long-term growth, enabled and augmented by machines.

This site exists so you can decide whether we should talk. Look around. If something resonates, reach out.

I love wide-ranging, deep conversations. Few things are more meaningful to me. I dislike small talk, defined simply as: "If you learn nothing new about the person after talking to them."

HOW I OPERATE & RELATE

In any situation, there are two things you can control: the attitude and the altitude from which you approach the problem.

My mornings tend to be tangential, creative, and all over the place. Reading, writing, synthesizing. It looks like dawdling from the outside, but it's contemplative time.

I'm an ambivert. I prefer in-person conversations, video, or long-form async email. Texting with me usually turns into novels.

I oscillate between clarity and overwhelm. I've learned my growth stems from learning to work and flow with that instead of fighting it.

I don't identify as an expert in anything. I'm a little bit good at a lot of things. I'm quicker to share my faults, than my upside.

I'm a loyal human, as evidenced by my long-term toxic relationship with the Cleveland Browns.

In the year before the iPhone, I thought I was going to be a musician or producer. I still enjoy playing and writing music when I can find the time.

WHAT I'M DOING NOW

Projects

Caring Ledger A daily commitment system with impact-based scoring that aligns app progress with real-world growth.
MCP Server / Knowledge Base Infrastructure that lets AI operate directly on a 50,000-word interconnected research knowledge base.
Essays / Substack Writing about agency, cognition, and mental health infrastructure to clarify thinking and communicate complex ideas.
agent-see.org A website for the AgentSee Open Lab, serving as a foundation for future community and research operations.

Caring Ledger

I believe structure and accountability are important.

I'm deeply interested in removing the friction and increasing the utilization of systems that aid in productivity. Most systems are exploitative or many existing tools optimize for the wrong things.

Simple in theory, hard in practice. Create incentives that align with the internal makeup of an individual, a system where progress in the app corresponds directly to real-world growth.

I'm using this system daily and continue to iterate on it.

I have built more complex systems in the past that used AI. At one time, I thought I was going to start a startup to do this very thing, and that led me down a different path The one that leads to me working on a human-AI co-regulation cognitive architecture.

I wrote zero code by hand.

Better to treat AI like a compiler for intent. Specify behavior, test it against reality, then refine the model.

I'm make so many mistakes everyday.

This is the fastest way to learn.

These days, it's a combination of Claude Code and Cursor.

Technically dead simple at this point:

Python/FastAPI backend, SQLite database, vanilla JS frontend.

I don't put every task or time box anything and everything. It's for things I'm committing to. Domains that feed my core loop that move me from fighting life to flowing through it: Read, Document/Process, Build, Outreach, Health, and Life.

The initial version of the system used time-based scoring.

After using it, I noticed that time-based scoring incentivized the wrong behavior. It rewarded spending time rather than producing meaningful outcomes.

One of the reasons I built it was that I struggle with time boxing.

Because of that, I changed the scoring model from time-based to impact-based.

This change came directly from observing my own behavior while using the system.

I want game progress to equal real-world growth.

If I can increase my score without becoming more capable or effective, the system is broken.

The system should be structured so that using it correctly makes me better, not just more active. This is fundamentally about incentive alignment: the easiest way to "win" should also be the healthiest and most effective way to behave.

That positive type of addiction is what I call "positive interdependence."

NOTE: The styling the concepts is heavily influenced by a book called "On Caring" by Milton Mayeroff. I recommend anyone and everyone read it.

MCP Server / Knowledge Base

I maintain a growing, interconnected research knowledge base spanning neuroscience, psychology, and systems thinking.

The knowledge base itself has been an emergent process started in March of 2025, formalizing my prior knowledge alongside new research and ongoing thinking in these areas.

Manually maintaining consistency across hundreds of documents is slow, error-prone, and not a good use of time.

So I built infrastructure to let AI operate directly on my files.

At this point, it exists as a book-length research corpus (approximately 50,000 words), organized as an interconnected knowledge graph rather than a linear manuscript.

24 documents.

38 internal links.

8 document types.

Semantic versioning.

Structured for both human navigation and programmatic maintenance.

See a likely outdated abridged version here: AgentSee Knowledge Base

I wasn't sure what kind of MCP server I needed at first, so utilizing the latest technology I built three versions in an 1 hour and selected the one that worked best.

The server allows Claude to read and edit markdown files on my local machine.

When I want to change how I define a term, update a concept, or reframe an idea, doing that manually across dozens or hundreds of files would take hours.

I would rather spend time thinking and building than doing repetitive find-and-replace work.

So I built automation to remove that friction.

Here's a quick demo of how I used it when writing this...

To create this webpage in literally 2 minutes

Essays / Substack

Against Declivity

I write essays about agency, cognition, and mental health infrastructure.

I write for two reasons:

To clarify and formalize my thinking about hard problems: AI, mental wellness, cognitive infrastructure, agency, and how to actually make progress.

To practice communicating science and complex ideas accessibly and share what I have learned.

I wish I was better and faster at this. It is a skill I am working on. This is where machines have been almost completely useless.

I'm going to try to write much shorter takes in 2026.

These essays were manually constructed only using AI for spell-check and grammar.

The earlier essays were more machine assisted, and I think they're crap.

If there's something you would like me to try and communicate, please let me know.

agent-see.org

AgentSee Open Lab

I designed and shipped a website for the AgentSee Open Lab.

(Honestly, several iterations.)

It's still woefully subpar from offering to execution (I'm aware, And will happily still accept feedback).

I'm still working on what this is going to become, but my aim is a foundation for a future community and research operation, rather than a one-off launch.

It is not dead, not correct. It is early.

WANT TO TALK?

This week I'm taking a lot of short calls to increase my feedback loop—if anything here intersects your world, grab time.

Book a call: Calendly link

DM on X: @JesseKowalcic

Email: jesseckowalcic@gmail.com

Reach out if you're building around cognition/agency + AI, or anything that smells like mental wellness infrastructure.

Reach out if you want help building something with AI, or you're stuck and want a clean read on what might be blocking you.

If you talk to yourself a lot, talk with me too.

I'm looking for aligned gigs—work that compounds what I'm doing.

I can build functional things fast with AI. I can do thoughtful research support.

My drivers are simple: I care more about the work being real and aligned than maximizing rate.

WHERE THIS IS GOING

We have more ways to service our cars than we do our minds. One of these is far more vital than the other.

Mental health has no real engineering discipline. Health care systems in the U.S. optimize for time-to-service, how fast someone gets help. They rarely measure time-to-regulation, how long it takes someone to return to a functional state.

I think much of human suffering is a systems failure, not a moral failure. Humans are unstable controllers running ancient biological code in modern environments. The load profile is incompatible with the organism. Evolution can't fix this in time.

Technology might.

I think there's a better future where we have cheap, efficient, machine-assisted co-regulation. Systems that help people see and steer their own state. That may not sound like much, until you find yourself being the right person in the wrong environment.

We need better infrastructure for human agency.