Closeout

Failure in. Proven fix out.

A factory failure usually closes on a yield bump. Then it comes back. Closeout closes it only when a held-out build proves the fix.

SubQ Snap 1.1 scores every failed unit against every cause. Muse Wonder patches the firmware. Gemini 4 Argon tries to break the root cause without seeing the fixer’s reasoning. Code runs the gates, and MongoDB Atlas keeps the issue list.

Watch the 60-second demo.

Results

• 20 held-out builds, 320 synthetic units. Cold failures went from 113 to 0.
• All 15 known-bad domes still fail. No warm unit broke.
• The shortcut, widening the optical limit, lets a known-bad dome through. The gate rejects it.
• Across 24 sealed scenarios: 1 teardown to root cause on average, against 1.08 for random picks and 3.71 for worst-first.

Current boundary

The product, the line and every unit are fictional and synthetic. The browser runs real compiled C. The SubQ call and the engine trace are recorded replays, labeled on screen. Scores rank causes; they aren’t calibrated probabilities. Closeout is a hackathon prototype, not an N+1 Labs product.

Run the bench or read the full field report.