Work with me · Fractional AI & automation leadership

Enterprise-grade AI,
on a few days a week.

For founders and operators of UK service businesses, roughly 10 to 100 staff, whose AI pilots never quite shipped.

Book the 2-week diagnostic

The whole deal is on this page, guarantees included.

Never run an AI pilot? Even better — you haven’t spent money learning what doesn’t work. The diagnostic starts from your jobs, quotes and invoices, not from the technology.

The consultation

What founders say.
What I say back.

Six objections from real first meetings. The last one is the deal.

  1. “Nothing ships.”
  2. “Nobody owns it.”
  3. “Is it any good?”
  4. “What's it worth?”
  5. “I'm not hiring a department.”
  6. “What about our data?”
  7. “What does it cost?”
“We've done three AI pilots. Everyone claps at the demo, then nothing ships.”
Founder
I don't build demos. I find the one workflow where AI pays for itself and put it into daily use — working every day, or I haven't done my job.
Charlie
“Everyone touched the AI thing, and now nobody owns it.”
Founder
Mine. I build it and I answer for the number it's meant to move — long after go-live.
Charlie
“Honestly? I've no idea if this thing is actually any good.”
Founder
I build the scoreboard first: your real cases, graded. “Good” becomes a number, and nothing ships unless it beats the last version.
Charlie
“Everyone says I need AI. Nobody can tell me what it's worth.”
Founder
Two weeks in your numbers finds the job that's quietly costing you most — and what fixing it pays back, against a real baseline, before you commit.
Charlie
“I'm not hiring a whole AI department on a maybe.”
Founder
One person does the whole job, from finding the lever to wiring the fix into how the business runs. You pay for output, not payroll.
Charlie
“The thought of AI loose in our customer data keeps me up at night.”
Founder
It only sees what it's allowed to see, and every answer shows its source. A person signs off anything that matters. That's how I already ship, under audit.
Charlie
“So what does it cost me to find out?”
Founder
Two weeks, fixed fee, agreed before we start. Nothing worth building? You don't pay.
Book the 2-week diagnostic →
Charlie

How an engagement runs.

Step 1 · Fixed fee

The 2-week diagnostic

  • Days 1–2access; shadow how the work actually runs
  • Days 3–6P&L walk-through; time the workflows that eat the week
  • Days 7–9shortlist to one lever; baseline it
  • Day 10written diagnosis, costed build plan, go / no-go

You keep the roadmap either way.

Step 2 · Monthly retainer

Embedded build

A few days a week inside your team, building the one system at that lever and measuring it against the diagnostic's baseline. Wound down when it runs itself.

Step 3 · Handover

It keeps working without me

Your team takes over a documented system — runbook, test suite, monitoring, access matrix — so nothing depends on me being in the building.

In writing, before any invoice

The diagnostic

If it doesn't surface a system worth building, the fee is waived.

Any build after it

One system live, its impact measured on the P&L, within 90 days — or we part ways.

Your first 90 days.

What hiring me actually buys.

Days 1–30 · Find the leak

The numbers before the technology: a diagnosis of the one bottleneck costing you most.

Days 31–60 · Build against it

One working system at that lever, live — the first hours handed back.

Days 61–90 · Make it self-governing

Safety rails and a rhythm of checks, so it keeps working after I stop watching it.

What that looked like on real engagements →

Questions founders ask.

Will this work in a business you don't already run?
Yes — that's what the diagnostic is for.It exists to learn an unfamiliar operation fast, and the fixed fee caps your risk while I do.
What does it cost?
A fixed fee, agreed before we start.You know the figure before you commit — deliberately below one stalled pilot or one wrong senior hire.
Will this disrupt my team or replace our systems?
No — I build on what you run.CRM, accounts and job system stay put. The first change your team notices is time coming back.
Isn't a smaller services business too small for this?
That's the sweet spot.You don't need a data team; you need one lever found and shipped. Ops-heavy businesses move the number fastest.
You run a company — are you actually available?
Yes — a small number of engagements, deliberately.No conflicts, and every engagement is documented and handed over properly.
We've been burned by consultants. How is this different?
A working system, not a deck.Its impact is measured on the P&L, and the costed roadmap is yours whether or not we carry on.
Self-taught — can I trust the depth?
Inspect the code.Everything I ship runs for years under external audit and monthly board review — and the testing system behind it is public: github.com/CharlieHulme/rag-eval-harness.

Different question? Email me →

Book the diagnostic.

One email is enough. I reply personally within a working day, with a fee and a start date.

Email me LinkedIn

Checking the tech for someone? The engineering →

github.com/CharlieHulme UK · remote / hybrid