adam hevenor
Independent AI and search consulting.
I help founders and technical leaders decide what to build, test the hard part, and ship AI and search systems. You work directly with me. I bring senior engineering judgment and the model-first tooling I use to build my own products.
the factory
I run an independent, bootstrapped practice. The factory is my in-house way of building: models do implementation work, tools carry context, and automated checks make the work reviewable. I set direction, review the result, and own the delivery.
Owned projects give that system real work: retrieval infrastructure, reproducible research, and a travel publishing business. Each has a different purpose and a different test of whether it works.
Consulting engagements and owned projects use the same working methods. Client case studies show the engagement; factory proof stories show what I build and measure on my own account. -> about Adam
case study · Trio Health A search api that became an agentic platform 
case study A two-week CLI sprint for an LLM evals platform 
case study · Taffrail An MCP server for a wealth-advisory rules engine r&d · active hev dots 
r&d · active hev layer 
r&d · active hev kit 
r&d · experiment hev ask 
showcase
A travel publication, built whole — destination guides, illustrated art direction, search, accounts. Next up: an agentic trip app builder.
-> travels-with-charlie.com
the client workspace
Built around weekly updates I write myself — what landed, what's next, what's blocked. The agent answers on top of them, grounded in the plan, the hours, and everything we produce. Yours to keep. Shown: an example week.
-> book a time
Workshops A hands-on half day with your team: operating an agent factory, evals from zero, KPIs that drive behavior, or one shared vocabulary across your leadership. Your team leaves with something working. 
Research Plan A hosted plan for the question you're actually trying to answer — what to learn, from whom, in what order, and what would change your mind. Your team marks progress against it and interrogates it with an agent. 
Research Program The plan, and I run it. Six interviews with people who can answer the question, then the insight analysis: what the market said, where it contradicts your roadmap, and what to do about that. 
Search & Retrieval POV A proof of value on your own corpus. Evals first, so "better" is a number you can defend, then hybrid, reranking, chunking, and the vector database call, argued from your data. You keep the eval set and the system. 
Fractional CTO I run a factory on your codebase. Agents do the volume, I own the review gate. Around that: architecture, vendor selection, hiring judgment, and delivery rhythm. 
Hourly Data engineering, ML, and strategy work billed by the hour — for a bounded problem, a second opinion, or ad-hoc help when you don't need a standing engagement. Top-quality reranking got about four fifths cheaper in a month. Perplexity's pplx-decider-v1.2 as a BEIR reranker: asked to choose among 30 documents it is statistically level with Mixedbread v3.1 listwise at 0.511 mean nDCG@10 for $0.20 per 1,000 queries from billed usage, and beats Jev. The flash model matches Jev at $0.10.
Why speed is the wrong reason to build a software factory: velocity metrics break once agents arrive, time to market is relative, and the outcomes that matter are quality, reliability and taste.
A search builder's read of the re:Invent 2026 catalog: vector search in every AWS database, agent retrieval replacing RAG, which vector databases are on the expo floor, and the sessions worth your time.
OpenAI's Decisions API (gpt-6-luna, public beta) as a BEIR reranker: 0.472 mean nDCG@10 in its best shape, behind Jev and Perplexity at 0.501, with peaked probabilities that break the 0.1 prune.