Proof index

Everything on this site is backed by something you can open. Here it all is, in one place.

#Public code

Clone it and run it yourself — these prove the method is real, not a slide. llm-eval-gate is a working CI eval-gate you can drop into a repo; playwright-sdet-regression-suite ships its evidence folder (traces, screenshots) alongside a 37/37 zero-flake run.

llm-eval-gate — keyless eval-gate template (MIT) →playwright-sdet-regression-suite — 37/37 in CI →GitHub profile →

#Verbatim runs & case studies

These prove the work survives contact with reality. The nexural-qa-os pair is the honest one to open first: the gate went red on a real CVE finding, blocked the release, and shows the green rerun after the fix — no quiet override.

Case studies — four measured outcomes →nexural-qa-os — the red run →nexural-qa-os — the green rerun →playwright suite — the run →

#Live & interactive

Proof you can operate yourself, right now. The live eval grades an AI in your browser — and its “bring your own AI” mode scores a real answer of yours against the same rubric I build for clients.

The live eval (grades an AI in real time) →I red-team my own AI (10/10 probes, static results) →Live demos (AI receptionist, voice, pipeline) →Sample eval report →ROI calculator →

▸

This documentation site is itself proof: it’s generated, static, and gated by its own Playwright + axe suite on every deploy.

Want your system on the case-studies page?

Start with a free mini-eval.

Build your plan → 2 min or book a call →

Was this helpful?Book a call →Get a free mini-eval →

← Previous

Glossary

Build an LLM Eval Gate in CI with Promptfoo