Automation
Every business runs on a set of small, dull, unavoidable steps. I find them, map them, and rebuild them as automations that keep running whether or not anyone remembers to press the button.
See what gets automatedSome automations sit behind your tools and simply run. Others sit in front of your customers and answer. Most businesses need a little of both, and the trick is knowing which is which.
A front desk that never closes. It answers the questions you get twenty times a week, pulls the real answer out of your systems, and hands over to a human the moment it is out of its depth.
Assistant
Hi — has my order shipped yet?
It left the warehouse this morning. Delivery is booked for Thursday between 1 and 5pm.
Can I move it to Friday?
Typing
Follow-up
Runs every weekday, 08:00
Tuesday's quote went unanswered, so a second note went out this morning and the thread is flagged for you.
Triage
Runs on arrival
Forty-one overnight emails sorted into quotes, invoices, suppliers and noise. Three of them need you.
Reviews
Runs the morning after a job closes
Job marked complete at 16:20 yesterday. The review link went out at 09:00 today, while it was still fresh.
Four places where the same work happens over and over. Pick the one that sounds like your week.
Web form, Instagram message, phone note, forwarded email — all of it arrives as one record with the same fields filled in, so nothing sits unread in a channel nobody checks.
The price list, the customer's details and the job type are already in your systems. The draft is assembled from them and waits for you to check it, rather than starting from a blank document at nine at night.
Unanswered quotes get a second and third nudge on a schedule you set, and stop the moment the customer replies.
Confirmation, calendar invite, directions and a reminder the day before — sent from your address, in your wording, without anyone typing them.
A polite reminder at seven days, a firmer one at fourteen, and a flag for you at twenty-eight. Silence is what costs money; the reminders cost nothing.
The same details typed into the CRM, then the accounts package, then the spreadsheet. One entry, written everywhere it needs to go, with no transcription mistakes.
Contracts, job sheets, certificates and handover packs generated from the record they belong to, named properly, and filed where you would have filed them.
New customer, new supplier or new starter: the folder, the welcome email and the tasks all appear in the right order without anyone remembering the order.
Opening hours, order status, lead times, what is included, how to reschedule. An assistant answers them from your own documents — not from guesswork — and says so plainly when it does not know.
When it hands over, it hands over everything: the conversation, the customer record and what has already been tried. No one has to ask the customer to start again.
Messages sorted by what they are and how urgent they are, so the one that matters is not third from the bottom.
The assistant reads from one source you control. Change the price list or the policy in that one place and every future answer changes with it.
Enquiries in, quotes out, jobs closed, invoices outstanding. Assembled overnight and short enough to read with a coffee.
Not a notification for everything. A message when something is genuinely off — an invoice past thirty days, a quote sitting untouched, a job with no engineer assigned.
Every figure in the summary points back to the records it came from. If a number looks wrong, you can open it and see exactly which rows produced it — which is the difference between a report and a rumour.
How the work goes
I sit with whoever actually does the work and write down every step, including the ones that only happen when something goes wrong. Most processes turn out to be three real decisions surrounded by twenty acts of copying. You get the map whether or not you build anything with me.
The first workflow goes live before the second is started, so you are getting value while the rest is still being written. Everything is built inside your own accounts, on the platforms you already have, with the failure cases handled: a run that cannot finish stops and tells someone instead of leaving half a record behind.
You get a written record of what runs, when it runs, what it touches and what to do if it stops — in plain English, not screenshots of a settings page. Logins and billing stay in your name from the first day, so nothing you depend on lives in an account you cannot reach.
Tools change their rules, price lists change, and a workflow written in March can quietly stop making sense by September. If you would rather not watch for that yourself, I keep an eye on the runs, fix what breaks, and adjust what has drifted. If you would rather own it entirely, the documentation is written so you can.
You do not have to automate the whole business. One process, done properly, pays for the work of finding the next one.
A call and a written map of your processes, with the automatable steps marked and the ones that should stay human marked too.
The single most repetitive thing you do, built end to end and live. Usually enquiries, invoicing or follow-up.
Three or four workflows that share data, so a customer's details are entered once and travel everywhere they are needed.
Monitoring, fixes and adjustments as the business changes. Optional, month to month, and easy to stop.
Work is quoted per scope after the audit, as a fixed price rather than an hourly rate.
The things people ask on the first call, answered before it.
Anything with a rule behind it. If a person is copying information between two screens, sending the same message with three details changed, or checking whether someone replied, that step can run on its own. Judgement calls — pricing an unusual job, handling an upset customer, deciding who to hire — stay with you, and should.
No. Automations are built on top of the tools you already pay for, using the connections those tools already offer. Replacing working software is the most expensive way to solve a workflow problem, and it usually solves a different one.
It stops, tells someone, and leaves your records as it found them rather than half written. Failure handling is part of every build, not an extra: an automation that fails loudly is safe, and one that fails quietly is worse than no automation at all.
Usually a matter of days once the process is mapped, because the mapping is the slow part and the building is not. Larger programmes are staged deliberately so something useful is running long before the whole plan is finished.
You do. Everything is built inside your own accounts, under your own logins and billing, and handed over with written documentation of what runs, when, and what to do if it stops. There is no arrangement where switching away means losing your own workflows.
Work is quoted per scope after a short audit call, as a fixed price rather than an hourly rate, so the number does not move while the work is being done. The audit tells you which processes are worth automating and which are not, and that answer is yours whether or not I build anything.
Start here
Tell me the process that eats your week. You will get a straight answer on whether it is worth automating — including when it is not.