Think inside your AI world.
The criteria layer
Stack the analytics layers and there is a gap at the bottom. The warehouse holds the data. The semantic layer holds the definitions. Above them, dashboards render it. Missing is the layer that holds the bar each number has to clear and the reason it sits where it does — the criteria layer. Without it, every read stops one step short of a decision.
A criterion is not a metric. A metric is gross margin is 61%. A criterion is margin must hold above 60% or the discount is pulled, because below 60% the blended CAC stops paying back. The first is data; the second is a decision with a reason. Unl is where the second kind lives, ratified, so a read is checked against it automatically — and the answer is a verdict, not a figure you still have to grade.
What sits in the criteria layer?
Three things, bound together per number: the bar (60% margin), the consequence if it is crossed (pull the discount), and the reason the bar is there (below it, CAC stops paying back). A metric alone carries none of these. A threshold written in a doc carries the first and loses the other two the moment the doc is closed. The criteria layer keeps all three attached, so a read can apply the bar and explain it in the same breath.
Crucially, what lives here is ratified, not guessed. The layer does not infer that margin should hold above 60% by pattern-matching similar companies; it holds the line you set, because a criterion inferred is a criterion nobody agreed to. When the read cites the bar, it is citing a decision on record, which is why the verdict is one you can stand behind in a room.
Why has no one built it before?
Because the criteria were never treated as data. They lived where decisions live — in a founder’s head, a kickoff deck, the memory of the meeting where the number was agreed. Analytics tooling assumed its job ended at rendering the metric and someone else’s job began at judging it. So the most decision-relevant information in the company — the bars and their reasons — was the one thing no system held.
Say you are running operations for a subscription brand. Your criteria are exact and everywhere in your reasoning: reorder when cover drops below six weeks, treat a cohort as healthy only above 40% month-two retention. None of that is in your analytics tool, because none of it is a metric — it is the set of standards you measure the metrics against, and until now there was nowhere for standards to live except you.
What changes once the layer exists?
The read gains a verdict. Your stock question stops returning “cover is 5.2 weeks” and starts returning “5.2 weeks — below your six-week reorder line, so reorder now; you set six weeks because the supplier lead time is four and you keep a fortnight’s buffer.” The metric came from the tool; the line, the action and the reason came from the layer that finally holds them.
And because the criteria are ratified and on record, each verdict is checkable after the fact: anyone can see which bar was applied and why it was the bar. That is the quiet payoff of building the missing layer — not only that reads become decisions, but that the decisions carry their own justification, ready to be audited rather than re-argued.
The criteria layer is the missing tier of the analytics stack: it holds, per number, the bar it must clear, the consequence of crossing it and the reason it sits there — ratified, not inferred — so a read returns a verdict you can stand behind instead of a figure you still have to grade.
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 is the criteria layer?
It is the tier that holds, for each number, the bar it must clear, the consequence if the bar is crossed, and the reason the bar sits where it does — ratified by you, applied at read time. It sits below the dashboard and beside the semantic layer, and it is what turns “cover is 5.2 weeks” into “below your six-week reorder line, reorder now, here’s why the line is six.”
How is a criterion different from a metric?
A metric is data — margin is 61%. A criterion is a decision with a reason — margin must hold above 60% or the discount is pulled, because below that CAC stops paying back. The metric describes; the criterion judges. Analytics has always rendered the first and left the second in someone’s head, which is exactly the gap the criteria layer closes.
Does it infer my thresholds automatically?
No — and deliberately so. A criterion inferred is a criterion nobody agreed to. The layer holds the line you ratified, so when a read cites the bar it is citing a decision on record, and the verdict is one you can stand behind. 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.
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.