Math gates: answers anyone can recompute

Some decisions should not depend on a model's mood. Does this agent pipeline have a cycle? Is this A/B result significant, and how many users do we need? Does this approval process pass the four-fifths rule? Three small services answer with plain arithmetic: same input, same answer, every time, and you can redo the math yourself.

#24 of the CAIN-42 Top 25 8 gaps found and fixed today shipped to the suite host 2026-11-02 19:02Z; live re-check pending

9/19

held on today's blind set, before any change

1,053

count pairs that float rounding wrongly flagged as unfair: now exact fractions

500 → 200

a 1,200-step pipeline crashed FlowGate live; now 20,000 steps pass

models in any decision

Try it

Check 8 on verify-yourself runs FlowGate, FairGate and LiftGate from your browser.

curl -s -X POST https://cainstudio.online/suite/flowgate/v1/validate -H 'Content-Type: application/json' \
  -d '{"tasks":[{"id":"plan","depends_on":["review"]},{"id":"review","depends_on":["plan"]}]}'

FlowGate · LiftGate · FairGate
Evidence: every blind row, before and after

FIRME AI CORPORATION · CAIN-42 is a Byzantine trust fabric. Numbers marked "our own check" were run by us, not by an independent party. Nothing on these pages is a certification.