Implementation
Find the bet worth making. Build the thing that makes it. Or hand you the plan.
Most AI consulting jumps straight to 'let's build a chatbot.' I don't. The biggest leverage usually shows up two questions earlier — what should we be building, and have we proven it works?
The engagement, step by step
Frame & Coach → Build & Teach → Read the Signal → Scale or Kill.
How a build runs end to end. I do the heavy engineering; your team learns the frameworks and tools as I build — so when the engagement ends, they own it and operate it, and the muscle stays.
01
Frame & Coach
Week 1. We map your business, your stack, your data — and write an opinionated thesis with two or three candidate bets. Your team learns the frameworks while we write it. The real work starts here, not later.
02
Build & Teach
Weeks 2–3. Real code, real prompts, real configs — not slideware. You learn the tools while we ship: prompts, context, agents, harnesses. Cost of building has collapsed; we use that, and you learn how.
03
Read the Signal
Week 4. Real users, real data. You learn to tell AI signal from noise. We size validation to risk — not to bill more hours.
04
Scale or Kill
Week 5+. Three paths: build it together, hand off a plan your team owns, or kill it and bank the savings. By now you can make the next call without us in the room. That's the point.
Why this method
Cost of building has collapsed.
A working AI prototype that took a team a quarter in 2023 takes a builder a week now. So we don't validate before building — we build to validate. Faster, cheaper, more honest.
Sovereignty by default.
Every artifact — code, prompts, library, configs, infra — lives in a repo you own. Vanilla tools (Claude Code, GitHub, your cloud) so you can swap models, harnesses, or vendors as the market evolves.
The exit is the offer.
We don't optimize for retainer expansion. The cleanest engagement is one that ends with you running.
Where we build
Your stack. Not ours.
One client runs Microsoft 365. We indexed 8,511 lease documents inside their own SharePoint. Another runs Google Workspace. We stood up the domain, the email and the shared Drive from scratch. We build where the work already lives.
SharePoint or Drive.
Your tenant, your access controls, your people’s logins. The corpus doesn’t move.
Outlook or Gmail.
Drafts land where you already work. Nothing sends without you.
No new platform.
Nothing to buy, nothing to upload somewhere else, nothing to migrate when we leave.
What you get
- → Written thesis document, branded with your context
- → Working prototype in a repo your team can poke at
- → Validation report with quantitative signal where possible
- → Scale plan: cost, time, team needs, vendor recommendations, build vs. buy
- → Decision log
Pricing
Thesis Sprint
$7,500–$18,000 fixed
Scope set on intro call.
Build & Hand-off
Scoped after Sprint
Typical engagements run $25,000–$95,000.
Plan-only hand-off
$0 additional
Included in Sprint deliverables. Your team builds, I'm available for spot consulting.
All deliverables are yours to keep. I bring the muscle. You keep the artifact.
Start with a Thesis Sprint.
Two weeks, fixed price, real artifact. By the end you'll know what to build, what to skip, and whether to use me to ship it.
Talk about a Sprint →