Knowledge Base · AI

AI shipped to production.

I don't strategize about AI — I deploy it. Claude-powered chat widgets, semantic knowledge bases, and agentic coding workflows, built end-to-end and running on real products. The fraud work pays the bills; this is where a growing share of my attention is going.

An honest note on this pillar. The fraud knowledge base runs deep — years of operator work. The AI pillar is younger and growing. What's below is what's live and real, not a padded catalogue. If you want to see where it's heading, the writing section is where I'll be building it out.

What I ship

end to end
Service
AI Automations for Growth Teams

Claude-powered chat widgets via edge functions (API keys stay server-side), semantic knowledge bases from existing support content, and SKILL.md authoring for agentic coding workflows. Branded, deployed, working on your stack.

 Proof
Case Study: AI Chat Widget Deployment

A help center turned into a production Claude experience on a Webflow site in under two weeks — one script tag, Vercel Edge, streaming responses with knowledge-base tool calls.

AI safety & trust

active interest
Background
Why fraud-fighting maps to AI safety

Years calibrating fraud filters is, at bottom, trust-and-safety work: separating real signal from adversarial noise without punishing legitimate users. That's the lens I bring to AI safety — and it's where I want more of my next decade to go.

 Reference depth
What a built-out pillar looks like

The fraud knowledge base shows the operator depth I am working toward on the AI side. AI writing builds out right here as I publish it.
Work with me

Want AI in production without hiring a team?

I take AI features from concept to deployed in a focused two-week sprint — and I bring the same engineering rigor I use on my own products.

See the engagement →