Think inside your AI world.
Should I kill this ad?
Every ad is a small standing decision: let it keep spending, or stop it. Kill too early and you throw away an ad that had not yet gathered enough data to prove itself; kill too late and you feed a loser for a week. The right moment is defined by a rule — enough spend to trust the signal, and a bar it then fails — and that rule is rarely written down.
“Should I kill this ad” is a timing decision disguised as a performance one. It needs two things you decide, not the platform: how much spend makes the result trustworthy, and the bar the result must clear once it is. Unl holds both, so the question returns kill-or-keep — and, crucially, “not yet, keep spending to the threshold” when the ad simply has not had its fair test.
Why is the timing so easy to get wrong?
Because raw ad performance is noisiest exactly when the decision feels most urgent. A new ad’s early numbers swing wildly on tiny samples, so a bad first day tempts a premature kill and a lucky first day tempts an over-investment — both reactions to noise the platform presents as signal. The instinct to act fast is strongest at the moment the data deserves it least.
So the decision needs a floor: enough spend or conversions that the result means something. Without that floor, kills are made on vibes and defended after the fact, and the same ad might be killed on Monday and, in another mood, kept on Tuesday. The platform shows the numbers; it does not hold your threshold for when to believe them.
What does the verdict weigh?
Mira, running paid acquisition, ratified a two-part stop-rule: no ad is judged until it has spent £150 — below that the sample is noise — and past that, it is killed if its cost per acquisition is above £40. Asked about a struggling ad, the read applies both parts: “keep for now — it’s spent £90, under your £150 judging threshold, so it hasn’t earned a verdict yet.” A week later: “kill — £160 spent, CPA £52, past your £40 bar.”
The £150 floor and the £40 bar are your decisions about trust and economics, and together they time the kill correctly. A model can see the ad is underperforming; it cannot say “keep, it hasn’t been tested yet,” because the judging threshold is yours, not the platform’s. The frame judges the data it is given; it does not verify the source’s accuracy.
What does the rule protect?
Both the budget and the learning. The spend floor stops good ads being killed before they prove out; the performance bar stops bad ads being fed on hope. Between them, the rule removes the two expensive errors that gut-timed kills produce, and it applies the same standard to every ad so the decision stops depending on the mood of the day.
And when you decide a higher CPA is acceptable for a prospecting ad than a retargeting one, you ratify the different bars, and each ad is judged against the one that fits its job. “Should I kill this ad” becomes a verdict against a rule that knows the difference between noise and a real miss.
Killing an ad is a timing decision that goes wrong because early performance is noisiest when the urge to act is strongest; measured context holds the ratified stop-rule — a spend floor for trust and a bar for performance — so the read returns kill, keep, or not-yet-tested rather than a nerve-driven guess.
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
Should I kill this underperforming ad?
Only if it’s had a fair test and then failed your bar. The decision needs two things you set: a spend floor that makes the result trustworthy, and the performance bar it must clear past that. Read against both, the answer can be ‘keep, only £90 spent against your £150 judging threshold’ or ‘kill, CPA £52 past your £40 bar’.
When should I pause an ad?
When it crosses your stop-rule, not when a bad day rattles you. Early ad numbers swing wildly on small samples, so a spend or conversion floor is what separates noise from signal — below it the ad hasn’t earned a verdict. Measured context applies that ratified floor and bar so the kill is timed on evidence, not nerve.
Can AI decide which ads to cut?
It can flag underperformers; it can’t time the kill, because the trust threshold and the performance bar are your decisions about noise and economics, not platform facts. Measured context holds them so the read returns kill, keep, or not-yet-tested. 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.