Quantified AI consulting

How I work

I start with a free 15-minute call, then an audit: interviews, a written report and a debrief. If you want me to build what the audit recommends, this is how that goes.

If you bring me back to build it

It's the same four steps every time, and you'll know what you're getting at the end of each one.

1
Interviews
Working sessions with your team to map how the work really happens today, where it gets stuck and what gets done twice.
You get: a shared, accurate picture of current state.
Consent-based interviews · transcript capture
2
The brief
I turn those conversations into a clear brief: prioritised opportunities, current- vs future-state, scope, and how we'll measure success, presented back for your sign-off before anything is built.
You get: an agreed brief and opportunity matrix.
Analysis · workflow mapping
3
The build
I deliver the solution: workflow standards, reusable prompt libraries, configured Claude Projects & skills, governance, and working assets, and a person checks everything before it's used.
You get: a working solution + clear documentation.
Claude · Cowork · Code
4
Handover
I train your team and AI champions and leave editable SOPs and a runbook, so you own and run it without me, then we track benefits against the baseline and prioritise what's next.
You get: a team that can run it without me, and before-and-after numbers to show what changed.
Training · SOPs · benefits tracker

How I handle your data

Here's what happens to anything you give me:

Conversations are recorded only with your explicit consent.
Transcripts and documents live in an agreed, access-controlled location.
Client data stays in approved, business-tier tools with model-training turned off.
Identifiers are de-identified where practical during drafting.
Material output is reviewed by a qualified human before it's used.
Your data is retained only as long as needed and deleted on request.

The principle behind every build

AI drafts, a human decides.
Automation removes the repetitive effort; accountability stays with your people.
Written once, used by everyone.
A good prompt or a standard gets written once and goes into a shared library, that way everybody's using it and it isn't stuck in one person's head.
You own it after handover.
By the time I hand over, you should be able to run it without me.

Grounded in systems I build and run

I run my own businesses on this (noomoo, CarbonCut, and one outreach engine that runs across them), so when I design a workflow for you I've a fair idea of what it takes to keep it running.