The maintenance problem
Hand-maintained context vs a substrate
There are two ways to keep an AI’s context current: groom a file by hand, or let a substrate hold your decisions and retire them as your thinking moves. The difference isn’t only effort saved — it’s where the currency comes from.
A file is only ever as current as your last edit, and it can’t tell you which lines you’ve let rot. A substrate makes being up to date structural: its unit is a decision with its reasoning, put there by a gesture and kept current until you supersede it. Currency stops being a task you perform and becomes a property of the store.
The hand-maintained file
You write it in advance, guessing what will matter. You edit it after each change, if you remember. You prune the stale lines by hand. You copy it per repo and per tool. It works, and every bit of the upkeep is on you. In fairness, for slow-moving conventions that upkeep is trivial — which is the case a file is genuinely made for.
Where hand-maintenance breaks
The upkeep scales with how fast your decisions move, and it always trails them. The file is current only up to your last edit, and it has no way to flag the lines that have expired. Grooming becomes unbounded work that competes with the work it’s meant to support — and the busier the project, the further behind the file falls.
What a substrate is
Not a bigger file. A substrate is a store whose unit is a decision-with-reasoning: entered by a gesture, made current until a later gesture supersedes it. The old call retires visibly rather than lingering; the reasoning rides along with each decision; and you stay in command of what’s held — saved on your word, superseded on your word. It’s a different shape, not a longer document.
Why the shape matters more than the labour
The win isn’t only less typing. A substrate can serve the right decision the moment it bears, and it won’t hand you a stale one dressed as current, because superseded means retired. A hand-groomed file does neither: it waits to be read, and it keeps whatever you last wrote, expired or not. Tuned for Claude, Claude Code, ChatGPT & Cursor at launch, extending across the AI ecosystem. Connects anywhere MCP does.
A hand-maintained file is only as current as your last edit; a substrate makes currency a property of the store — decisions held on your word and retired on your word — so the answer comes back measured against what you already decided, and why.
Reads through Unl arrive with measured context — in the presence of the decisions you’ve already settled. The reach lane is live: one box, paste anything. If it speaks MCP, Unl can reach it. Readings arrive unprompted, the data beside the criterion; Unl is a courier, not a warehouse, and keeps only your keys and the frame.
Read further
Questions people ask
What's wrong with just maintaining a context file?
Nothing, for slow-moving conventions. The cost appears when your decisions move faster than you edit: the file trails your thinking, keeps stale lines that look current, and has to be groomed by hand forever. The upkeep competes with the work it's meant to support.
What do you mean by a 'substrate'?
A store whose unit is a decision with its reasoning, put there by a deliberate gesture and kept current until you supersede it. It's not a longer document — it's a different shape, where being up to date is built in rather than something you perform.
Does a substrate mean the AI decides what to keep?
No — you do. A decision is held because you said keep it, and retired because you superseded it. The substrate tracks currency and reasoning; the gestures are yours, so nothing lands or leaves without your word.
Do I still need my files?
Yes — keep them for conventions, commands and reference. The substrate carries the moving half: the decisions and their why. They compose — the file holds what's stable, the substrate holds what's current.
What this is
Think inside your AI world — you stay in command
Unlimitless (Unl to friends) holds what you've settled, reads what your tools are showing, and catches what's changed out in the world — and hands your AI whatever bears on the work, the moment it's needed, without you asking. The right thing, in front of the model, unprompted, with you in command of the call. So you keep moving toward what you set out to build, on top of everything you've already decided.
It plugs into Claude, Claude Code, ChatGPT and Cursor as an MCP connector. Quick to connect, in a couple of steps.
Unlimitless is open now to invited Alpha. Apply for the Beta waitlist to come in ahead of the full launch:
Alpha is invite-only · free at launch.