Book a call

For businesses of 10 to 200 people.

Where two or three people spend their week moving information between systems by hand.

Get your team's week back. In 2 weeks, or you don't pay.

The Bottleneck Sprint. I find the one job your team does by hand that eats the most hours, build the thing that does it instead, and prove it runs on your real data before you pay a penny. Capacity without hiring anyone.

What you get

One sprint, everything below, one fixed price. No day rate, no change requests, no scope creep.

If bought separately£30,200
Your price£7,500

A contractor at a normal day rate bills five weeks for this, with no fixed price and no guarantee. A consultancy costs more and starts next quarter. But the number that matters isn't the price: it's what another year of 2,040 hours costs you.

What it costs your team

The real question with any internal project is how much of your people's time it eats. Every one you have bought before consumed months of somebody's Tuesdays. This one is built so it doesn't.

2 hrsShowing me the process
1 hrGiving me access
1 hrSigning it off

4 hours of one person's time, total. No steering group, no weekly status call, no project manager. I work from your systems and come back when it runs.

The test that decides it

Before any money changes hands we write down one sentence that defines done. Not a scope document. One sentence, in your words, that is either true or false at the end.

Pass or refund

The process runs end to end, on your real data, for five working days, without anyone on your team touching it. If it doesn't, you get a full refund and you keep everything I built. The risk is mine, not yours.

This is the whole difference. Everyone else in this market sells days and slide decks, and you carry the risk that it doesn't work. Here the seller carries it.

What this looks like in practice

Here is what the arithmetic looks like. The numbers below are modelled rather than a past client, but the shape is the same in a bookkeeping practice, a recruitment firm, a broker or a builder's merchant: two or three people, one repeated task, and a total nobody has ever added up.

Before, per year · Modelled example, 22-person practice
LineHoursCost
Chasing clients for records920£27,600
Sorting receipts into the ledger640£19,200
VAT return preparation, quarterly480£19,200
Manual cost, per year2,040£66,000
After · Modelled example, 22-person practice
LineValue
Build, one fixed price£7,500
Staff hours returned per year1,800
Records chased to ledger-ready9 days to same day
Pays for itself in6 weeks

The acceptance test for a job like this: one full cycle processed end to end, output signed off by your team as equal to what they produce by hand, with no manual step in between.

How it runs

Week 0

Audit

Two days finding where the hours actually go. You get the map and a fixed quote whether or not you go further.

Weeks 1 to 2

Build

I build against your real systems. You see it working before the end of week one.

Week 3

Prove it

Five working days running untouched. It passes, or you are refunded and keep the work.

Capacity

2 sprints at a time. I do the work myself, so when both are running the next start date is the following month.

Start smaller if you'd rather

What I ask on the first call

  1. 01What does someone on your team do every week that they hate?
  2. 02How many hours is that, across how many people?
  3. 03What does it cost you when it is done late, or done wrong?
  4. 04Does the information already live somewhere a computer can reach?
  5. 05If it ran itself from Monday, what would that person do instead?

If the answers don't add up to more than the build costs, I'll say so on the call and we stop there.

Fit

Works when

  • You have ten to two hundred staff and real revenue.
  • Someone can name the process in a single sentence.
  • The data lives in software, not only on paper or in someone's head.
  • One person can approve five figures without a committee.

Doesn't work when

  • You want an AI strategy rather than one thing fixed.
  • Nobody can point at a specific process.
  • The goal is cutting headcount rather than freeing up hours.
  • Nothing bad happens if this waits another year.

Why me and not a consultancy

Most people selling AI to businesses right now cannot deploy software. They run the workshop, write the strategy and hand you a document. The build goes to someone else, or nowhere.

I'm an engineer, based in Ebbsfleet. I build production systems where being wrong has consequences, including autonomous navigation software and the hardware-in-the-loop rigs used to test it. You are hiring the person who builds it, not the person who scopes it. That is also why I can put a refund behind the outcome: I'm carrying the delivery risk either way.