✓ coming soon · the reusable IP pack
AI QA in a Box
The judgment layer, packaged. Everything I install in a Foundation Sprint (the rubrics, scaffolds, checklists, and playbook), as a pack you can run yourself.
$300 USD one-time · both TS + Python editions
The free starter is on GitHub now. The paid pack lands at launch (payments aren’t wired yet). Drop your email via the contact form to hear when it ships, or have it installed instead.
✓ what's inside
Five artifacts. The same ones I deliver in a sprint.
Authored net-new, generic patterns, no client-specific anything. The taste is the product, these encode it so a team can self-serve the parts they’re ready for.
- Eval rubric packs: ready-to-adapt scoring rubrics for common AI features (support replies, summarization, RAG answers, classification), with example golden cases.
- Prompt regression harness scaffold: a starter repo structure that compares prompt changes against a baseline and fails CI on regression.
- Injection threat-model checklist: the attack classes to test, how to interpret results, and remediation patterns (not just a scanner run).
- Cost-monitoring dashboard spec: what to instrument for API cost and latency, the guardrail thresholds, and how to fail a build on a cost regression.
- “First 90 Days of QA at an AI Startup” playbook: the sequenced plan: what to build week by week, what to skip, and how to install ownership.
✓ who it's for
For teams who want to DIY, for now.
If a $15k sprint is probably overkill for now but you’re past shipping on vibes, the pack gets you the structure without the engagement. It’s also a head start for the QA engineer you’ll eventually hire.
When you want it installed, tuned to your stack, and owned, that’s the Foundation Sprint. The pack is the same judgment, self-served.
Want a head start before launch?
The free grader gives you a scored roadmap today, and the contact form is where to register interest in the pack.