selective consulting
Bring me the product problem that is resisting polite conversation.
As a forward-deployed AI lead, I take a small number of MidnightDev engagements where product judgment, AI behavior, production engineering, and distribution have to be solved together. Scope and pricing follow the evidence—not a theatrical menu of agency packages.
how an engagement begins
Start with the smallest engagement that can change the decision.
The goal is not to manufacture a six-month roadmap. It is to expose the highest-risk assumption, test it honestly, and earn the right to do more.
01
Product diagnostic
A focused examination of the thesis, user loop, architecture, model behavior, acquisition path, unit economics, evidence, and failure modes.
You leave with a blunt decision memo: what is true, what is assumed, what to test first, and what should not be built.
02
Proof-of-value build
A sharply bounded implementation designed to validate the hardest technical or commercial assumption in production conditions.
You leave with a working system, measurement plan, documented tradeoffs, and an explicit build, change, or kill gate.
Product thesis and business design
Interrogate the opportunity, user loop, monetization, operating model, assumptions, and kill criteria before expensive certainty theater begins.
AI architecture and model behavior
Turn model capability into a reliable product with persistent behavior, explicit authority boundaries, human review, evaluations, fallbacks, analytics, and cost controls.
Organic discovery and distribution
Build the technical SEO, content architecture, programmatic search, GEO/AI-answer readiness, App Store positioning, owned media, and PostHog measurement system around the product.
Technical diligence and product rescue
Audit AI-built products, forecasts, and codebases for false assumptions, security gaps, brittle state, hidden cost, unsupported claims, and missing release controls.
AI-supported operating systems
Create shared context, decision records, agent workflows, PostHog dashboards, verification gates, and accountability systems that give small teams leverage without losing the plot.
Focused full-stack builds
Ship bounded web and mobile products across React, Next.js, TypeScript, Python, Supabase, Vercel, and Capacitor—with the business and measurement systems attached.
start with the truth
Tell me what must work, what already exists, and what has failed.
I'll tell you plainly whether I can help, what I would test first, and whether the idea deserves more capital.
← back to the workfolio