The Day the Counter Moved: Closed Loops 0 → 2
For weeks our closed-loop counter sat at zero on purpose — a promise we hadn't earned yet. On one day in August it moved to two: one bet graded a miss, one graded a hit. Here is why that day mattered more than either result.
Published August 13, 2026
Field note · events of August 7, 2026 · published August 13, 2026. The dates are real and checkable — the roundups and the ledger entries behind this note are dated primary sources, which is rather the point.
For weeks, one number on our own scoreboard sat at zero, and we left it there. Closed loops: 0. It was the most honest number on the board and the least flattering, and every time someone suggested nudging it, the answer was the same: a loop is not closed until its verdict is measured against the prediction it made before it ran. We had plenty of activity. We had, by the strict definition, nothing yet proven.
Then it moved
On one day in August it went to two. Not because we relaxed the definition — because two bets actually finished.
The first was a non-brand search experiment we killed. It spent roughly six hundred dollars, missed the cost-per-outcome it had promised at birth, and was shut off on the evidence. A miss. Graded, dated, kept.
The second was a marketplace coupon that beat its baseline by a wide margin and held a real share of the lift after the discount ended. A hit. Same machine, same week, opposite result.
Why zero-to-two beats zero-to-one
If the counter had moved to one, we’d have shown you a win, and you’d have been right to squint. One receipt is an anecdote, and anecdotes are what everyone in this category is selling. Zero to two is different, because one of the two is a loss we volunteered. A scoreboard that can land on the wrong side of the line — and shows it when it does — is a scoreboard you can believe when it lands on the right side.
That is the whole thesis of this company in one day’s data: not “our agents win,” but “here is one bet that paid and one that didn’t, graded by the same rules, and here is exactly how we knew.” The number that matters is not two. It is that two was arrived at honestly, on an append-only record nobody can quietly edit later.
What’s next for the counter
Up, slowly, and with more losses in the mix than a demo would ever show you. Every new loop is one more entry in the outcome ledger, and the ratio of hits to misses over time is the thing that turns running agents into proven ones. We’ll keep publishing both. The misses are the receipts that make the hits worth anything.
Questions founders ask
- What is a closed loop, and why count them?
- A closed loop is an action that was born with a prediction and has since received its measured verdict — the full arc from intent to outcome, recorded and graded. We count closed loops because it is the one number that cannot be faked by activity: an agent can take a thousand actions and still have zero closed loops if none of them were measured against a prediction made before they ran.
- Why was the counter honestly at zero for so long?
- Because measuring outcomes against pre-registered predictions is slow and unglamorous, and we refused to inflate the number with actions that were never actually graded. Zero was the honest reading while the loops were still open. Moving it to two meant two bets had genuinely closed — one that worked and one that didn't — not that we had gotten better at counting.