Some call this a forward deployed engineer: an engineer who sits with your team, learns how the work really runs, and builds the fix in the tools you already use. I call it engineering. Map, sort, build, prove.
I spent 22 years in engineering, the last four as Manager of Engineering for a 70-person department. Twenty years ago, a large part of that work was standardizing equipment and whole production lines. Find what repeats. Agree on one way to do it. Stop designing it from scratch every time.
It saved a lot of time, and it never started with a new tool. It started with watching how the work was really done. That was a long time ago, and I do not claim it as recent work. It is where I learned to solve problems this way.
AI changed what I can build. It did not change the method. Since 2014 I have used it on my own two brands first, and now on other people’s businesses.
I am early in doing this for other companies, and I will say so. What I can show you is my own work: real problems, in businesses and an event I run, solved with systems I built.
Ask anyone how a job gets done and you hear the short version. Watch the job, and read what the inbox and the systems recorded, and you find the rest.
Chasing, waiting and doing it twice. Nobody lists these steps, and they are where the days go.
An illustration, not a client result. Your map is drawn from your own work.
I sit with the people who do the work and watch them do it. Then I read what your inbox and your systems recorded. People give me the why. The records give me the what.
A job that takes ten days often holds twenty minutes of work. The rest is waiting on someone. I measure both, because the waiting is where the fix is.
One person has to be able to say yes to the change. We name that person before anything is built. This step has nothing to do with AI.
Delete it, give it to plain code, give it to an AI agent, or keep it with a person. In that order.
Before I build, we record how the process runs today: steps, days, how often work goes backwards, hands-on minutes. That is the mark I am held to.
I build inside the tools you already use. At 30 and 90 days I measure against the same numbers. If it did not move, you will see that too.
Most people start with AI. I start by deleting. Many steps never need AI at all.
Some steps exist only because an earlier step went wrong. Fix the cause and the step is gone.
For exampleChasing a missing detail. Typing the same order twice. Checking the same thing a second time.
If the rule is “when this happens, do that”, it needs no AI. Code is cheaper and never guesses.
For exampleOrder paid, send the invoice. New booking, add it to the calendar.
Where the work takes reading and judgement, an agent prepares it. Then it stops and asks.
For exampleRead the email, pull out the details, draft the reply for you to approve.
Anything you cannot afford to get wrong stays with you or your team.
For exampleApproving a refund. Sending money. Signing a quote. Talking to the client.
I do not ask you to replace your accounting system, your store or your booking system. The fix runs inside them. Your team keeps working where they work today.
When a decision needs a person, it arrives as an email or a Slack message with the facts laid out and a clear yes or no. The AI prepares the work, then stops. A person decides.
You own all of it: the code, the accounts, the data. No licence, no lock-in.
Clinics are where I most want to help. I run two national clinic directories, and I am taking on my first clinic process projects now.
I have run two brands on Amazon and Shopify since 2014. I built Helm for this work, including tracking reimbursement claims.
Accounting, legal, property and trades offices share one problem: chasing people for documents.
I am Race Director of the Bronte Harbour Classic and built the system behind it.
We find the one place where work piles up. You leave with a first step you can take without me.
I watch the work, talk to the people who do it, and read what the systems recorded. Read-only. I change nothing.
Every step sorted into its place. Today’s numbers recorded. A written build plan with real tools and real costs.
Deleted steps first, then plain rules, then AI where it earns its place. It runs beside your team before it takes over.
Same numbers, measured again. Documentation and training, so your team runs it without me.
One bottleneck, one tool named, one first step.
One process mapped and sorted, with a written build plan.
A whole department over two to three weeks: interviews, a read-only look at your systems, and today’s numbers written down.
Built inside your tools, measured at 30 and 90 days, handed over with training.