The mechanism
The frozen generalist
A frontier model is a frozen generalist: its weights don’t change after training, and out of the box it knows a little about everything and nothing about your project. It performs like a specialist only when the right depth of your specifics arrives in its context at the moment it’s reasoning.
The model is fixed; what varies is what reaches it. Give a frozen generalist the shallow version of your situation and you get a competent generic answer. Give it the deep, specific, current version — the decisions you settled and why — at the moment it bears, and the same model reasons like someone who has worked on your project for months. Unl is built to make that last delivery happen on time.
Frozen means frozen
Training ends and the weights freeze. Every session you meet the same generalist; it doesn’t learn you between chats, and it starts each one from the same trained baseline. That’s not a flaw to lament — it’s the shape of the thing, and it’s precisely why what you feed it matters so much.
Generalist by default, specialist on demand
The model already holds the latent ability to reason like a specialist in your project; what it lacks is your project. Specialism isn’t in the weights — it’s in the context you supply. The same frozen model is generic or expert depending entirely on what reaches it, which is a hopeful fact: you don’t need a better model, you need the right specifics in front of the one you have.
The right depth, at the right moment
Two failure modes bracket the target. Too little, and you get a generic answer. Too much at the wrong time — your whole history pasted in — and the one decision that bears is buried in the pile. Specialist behaviour needs the specific call relevant to this turn, delivered at this turn: depth and timing together, not an archive and not a guess.
Why this is the appreciating layer
Models commoditise; connectors commoditise. What turns a frozen generalist into your specialist is your ratified judgement, delivered when it bears — and that delivery keeps its value as the model underneath gets cheaper and better. The scarce thing isn’t the intelligence; it’s the specific, current judgement reaching it at the right moment.
What delivering it looks like
Unl holds your decisions with their reasoning and serves the one that bears as unprompted relevant context, so the frozen generalist reasons from your settled position rather than a generic default. What reaches it is only ever what you chose to keep and haven’t yet replaced — the specifics stay yours to set. Tuned for Claude, Claude Code, ChatGPT & Cursor at launch, extending across the AI ecosystem. Connects anywhere MCP does.
A frozen generalist becomes your specialist only when the right depth of your specifics reaches it at the right moment; supplying that — the decision that bears, with its why, exactly when it’s reasoning — is the whole game, 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 does 'frozen generalist' mean?
A frontier model's weights are fixed after training, so it's a generalist that knows a little about everything and nothing specific to your project. It doesn't learn you between sessions; each chat starts from the same trained baseline.
If the model is frozen, how does it act like a specialist?
Through context. The ability to reason like a specialist is latent in the model; what supplies the specialism is what reaches its context — your specific, current decisions. Deliver those at the moment they bear and the same frozen model reasons like an expert on your project.
Why does the timing matter, not just the information?
Because too much at the wrong moment buries the one thing that bears. Specialist behaviour needs the specific decision relevant to this turn, delivered at this turn — not your whole history pasted in, where the call that matters is lost in the pile.
Isn't a bigger context window the answer?
A larger window lets you paste more, but more isn't the same as the right thing at the right moment. What turns the generalist into your specialist is the precise decision that bears, with its reasoning, served when it's needed — which is what Unl is built to do.
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.