Review checklist · public-safe

Robotics falsification checklist

The right question is not which robot looks best. It is which claim survives falsification: production, deployment, reliability, customer economics and financial materiality.

Draft updated 2026-06-13. This page is a working surface for review before the final publishing flow.

Anti-hype gates

Designed capacity is not production

Tesla disclosed designed capacity and long-term designed annual capacity; actual output, accepted units, yield, utilization and economics remain separate questions.

Gate 1

Customer KPI is not full economics

Figure has stronger deployment evidence than demos, but S5 still requires customer-confirmed ROI/payback, repeatability, intervention data, service cost and gross margin.

Gate 2

Low price is not reliability

Unitree price anchors widen access and experimentation; they do not by themselves prove industrial uptime, support burden, margin, shipments or customer payback.

Gate 3

Supplier filings are not named humanoid exposure

Leaderdrive filings are stronger than narrative, but public copy should not infer unnamed humanoid OEM customers or humanoid-specific margins without primary evidence.

Gate 4

Commercialization remains testable

The honest framing is S4-heavy evidence: better than demo-only, still short of repeatable scaled commercial economics.

Conclusion

Public-safe wording rule

Use evidence-map language. Do not turn capacity, customer logos, low price, financing or supplier filings into a winner ranking, a trade recommendation or a claim that scaled commercial economics have already been proven.

Sources