About Nibbin

Most AI tools hand you a blank box and wish you luck.

The models are good now. That stopped being the problem a while ago. The real gap is between what a model can do and what you can actually get it to do. That gap is wider than anyone building AI wants to admit.

To get real work out of these tools, you have to know what to ask, how to phrase it, what context to paste in, and how to break your own process into steps a machine can follow. That’s a skill. Most people don’t have it, and there’s no reason they should. Their job was never “prompt engineer.” So the capability sits there, mostly unused, while people keep doing by hand the exact things the model could already handle, if only someone had set it up right.

We’re a small team that has spent about fifteen years building payments, risk, and AI systems at Google, Block, and PayPal, for products used by tens of millions of people, in regulated places where trust has to be earned. The capability in these tools is real. Most people just never get a fair shot at it, and that’s the gap we’re building Nibbin to close.

The part nobody fixes

Every tool in this space asks you to do the same thing first: describe your work.

Chat tools start you on a blank page. Every session, you re-explain who you are and what you’re working on before you get anything useful. Automation builders make you map out each step by hand before a single thing runs. Agent platforms want you to write a spec for an agent you haven’t even tested yet.

They all share one assumption: that you can explain your work well enough to configure a machine to do it. But the reason you wanted help in the first place is that your work is messy. It’s full of judgment calls, exceptions, and small decisions you make without thinking about them. Ask someone to write down their process and you get a clean version that leaves out everything that actually matters. Watch them do it and you see what really happens.

That’s the whole thing. The description is the lie. The work is the truth.

What it asks of youWhere it breaks
Chat toolsWhat it asksStart over and re-explain everything each sessionWhere it breaksNo memory of your actual work; you’re the context engine
Automation buildersWhat it asksMap every step by hand before anything runsWhere it breaksYou have to describe a process you’ve never written down
Agent platformsWhat it asksSpec out an agent up frontWhere it breaksYou’re guessing at a workflow you haven’t tested
NibbinWhat it asksNothing extra, just do your workWhy it worksIt learns from what actually happens, not your description of it

What Nibbin does instead

Nibbin learns from the work, not from your account of it.

It runs on your machine and pays attention to how you actually get things done: the real sequence, the real files, the real choices you make. From that, it builds agents and workflows out of what it sees. You don’t write prompts. You don’t draw flowcharts. You don’t sit down for an afternoon to “set up your automations.” You do your work, like always, and Nibbin turns the parts you keep repeating into Nibbins that can take them off your plate.

It stays local and private by default. The observer watching your work is yours, on your device, working only for you.

This might sound familiar

The weekly thing you’ve never explained to anyone.

There’s some task you do every week that lives entirely in your hands. You’ve never written it down because you’ve never needed to. A chat tool can’t help with it: you’d have to explain it first, and explaining it is half the work. Nibbin just watches you do it and learns the steps.

The automation you’d build if you had a free afternoon.

You know exactly which part of your week is wasteful. You also know that sitting down to wire it up in some builder would cost more time than it saves this month, this quarter, maybe this year. So it never happens. Nibbin removes the afternoon. The setup is you doing the work.

The context you paste in every single time.

Every chat session, you re-establish the same background before you get anything done. Who you are, what the project is, how you like things handled. Nibbin already knows, because it was there when the work happened.

We’re building Nibbin now.

The waitlist is open. If a tool that learns your work instead of making you explain it sounds like what you’ve been waiting for, come get early access.

Join the waitlist