Services
Three ways to work together.
Prove it, ship it, or run it. Pick the shape that matches where you are — each one is a fixed step with something you can hold at the end.
Prove it
Discovery Sprint
One to two weeks
A working prototype and a written technical plan.
Best for: An idea that needs to be proven before anyone commits a budget.
Ship it
Fixed-Scope Build
Four to twelve weeks
A production application, delivered and handed over.
Best for: A defined problem with a known outcome.
Run it
Fractional CTO
Ongoing, monthly
Senior technology leadership without a senior technology salary.
Best for: Companies past the point where a technology strategy can live in someone's head.
How I engage
You always know what you are getting, what it costs, and when it lands.
No open-ended hourly billing and no discovery that never ends. Every engagement starts with a fixed first step and every step produces something you can actually hold — a document, a working URL, a deployed app, a runbook. That includes being straight with you about where AI is doing the work and where a person still has to.
- 01
Week 1
Scope
One working session with your team. We define the outcome, the data sources, and what done looks like — before anyone commits a budget.
DELIVERABLE · Written scope + architecture sketch - 02
Weeks 1–2
Prototype
A clickable prototype running on your real data, not a slide deck. You see the thing before you fund the thing.
DELIVERABLE · Live prototype URL + effort estimate - 03
Weeks 3–12
Build
Production application: design system, auth and roles, data integration, security review, deployment.
DELIVERABLE · Deployed application + test evidence - 04
Final week
Hand off
Documentation, a runbook, and your team owning it. No hostage code and no dependency on me to keep the lights on.
DELIVERABLE · Runbook + handover session + 30 days support
How AI shows up in your engagement — and where it stops.
Where AI earns its place
Delivery speed
AI-assisted development is why one person now ships what used to take a team of six. You pay for the outcome, not the headcount.
Document and data extraction
Turning PDFs, certificates, and supplier paperwork into structured records a database can actually query.
Drafting and scaffolding
First-pass code, migrations, and test fixtures — reviewed line by line before anything reaches your environment.
Pattern-finding in your own data
Anomalies, outliers, and correlations surfaced for a human to judge. Surfaced, not decided.
Where it stops
Your data does not train anyone's model
Extraction runs against scoped credentials with retention off, and that goes in the contract, not just the conversation.
No AI in the decision path without a human
Anything that moves money, releases product, or affects a person gets reviewed by someone accountable.
No black-box numbers
Every figure on a dashboard traces to a defined measure with a named owner. If a model produced it, the derivation is visible.
No model where a query would do
Most ‘AI problems’ in an operation are a data-model problem wearing a costume. I will tell you when that is the case, even though the honest answer is the cheaper one.
Not sure which one fits?
If AI is the right answer, I’ll show you where. If it isn’t, I’ll tell you that too.
Describe the problem. I'll tell you which engagement it is — or that you don't need one.