What if we want you to actually ship to production, not just hand off a prototype?
That’s the default. The engagement ships to production, not a prototype. By week 6 your engineer has written the integration code, with me pairing and designing the evals throughout, and it runs in your stack. Operating it long-term (monitoring, on-call, evolving evals as models change) is a separate 90-day retainer if you want it; most teams run it themselves.
Which department should we pick?
We decide together on the first call. The strongest opportunities I’ve seen lately: legal (contract intake, NDA review), finance (invoice processing, AP, close acceleration), customer success (renewal prep, account health), support (deflection, agent assist), sales ops (lead qualification, CRM enrichment), and HR (résumé screening, help-desk). Have one in mind? We’ll pressure-test the fit. If not, I’ll help you pick.
What AI models and tools do you work with?
Primarily Claude and the OpenAI frontier models, built in Cursor. I’m model-agnostic in principle and use whatever best fits your data, latency, cost, and privacy needs, including within a cloud you’re committed to (Azure OpenAI, Bedrock, Vertex).
How do you handle our data and IP?
Mutual NDA before discovery. I don’t train on your data or reuse it across clients, and the feature plus every deliverable is yours. Specific compliance needs (SOC 2, HIPAA, GDPR residency)? Raise them on the first call and we’ll scope what’s possible.
What do you need from our engineering team?
One engineer, senior enough to make integration calls in your stack, with time freed across weeks 2 to 6. They write the production code. They have the access and context, and I don’t need it. I bring the AI product management you don’t have yet: what to build, the evals that prove it works, and the AI-specific calls. I pair with them daily; by week 6 they own the codebase and can maintain it without me. No prior LLM experience required.
Can the same engagement cover a second department?
No. One engagement, one department. For more, we run separate engagements, sequential or parallel. Many clients do two or three over six months.
Can the engagement be shorter or longer than six weeks?
Six weeks is the box: long enough for real work, short enough to keep momentum. Longer invites scope creep; shorter doesn’t produce something shippable.
Why fixed fee instead of hourly?
I get faster every time I do this, and hourly would punish that. Fixed fee aligns us: a known outcome at a known price, and neither of us watching the clock.
What if it turns out my workflow can’t be built in my environment?
That’s what Phase 1 is for. Discovery and system design happen before the build clock starts. If they surface a blocker (missing infrastructure, inaccessible data, a security gate), you find out at the checkpoint, where either of us can walk. You also leave with a written list of anything to provision.
Do you offer team training as a separate engagement?
Yes. Some clients want a dedicated training day for their PM and engineering leads on scoping, building, and managing AI features going forward. It’s a separate engagement, priced by team size and depth. Ask on the call.
Where are you based, and where will you work?
The San Francisco Bay Area. The engagement runs async-first over video; I don’t travel to client offices unless there’s a specific reason, billed at cost if so.
How quickly can you start?
Usually two to four weeks out. I run one or two engagements at a time, so availability rotates. Check current timing on the first call.
Still have a question? Book a 20-minute fit call and ask it directly.