Automation
Removing the work nobody should be doing by hand. Not a transformation programme — usually one specific afternoon-a-week task that shouldn’t exist.
Most automation worth building is unglamorous. Someone copies figures from one system into another every Monday. Orders arrive in one format and have to be re-keyed into another. A report gets assembled by hand from four places when it could assemble itself.
We look for the task you’d describe with a sigh, work out whether it can be removed rather than sped up, and build the smallest thing that removes it. If a spreadsheet formula or a setting in software you already own would do it, we’ll tell you that instead — it’s a shorter conversation but a more useful one.
Fit
When this is the right job — and when it isn’t
We’d rather lose a project at this stage than sell you one you don’t need. The right-hand column is the honest half.
- The same data gets typed into two different systems by a person, every week
- A recurring report is assembled by hand and takes half a day
- Orders, bookings or leads arrive in one place and have to be moved somewhere else manually
- Something important gets missed occasionally because it depends on someone remembering
- The process changes every few weeks. Automate it once it’s settled, or the automation spends its life catching up
- It happens twice a year. The work to automate it outweighs the hours it saves, and we’d rather say so
- The tool you already pay for does this and it isn’t switched on. That happens more than you’d think
Scope
What’s included
The audit first
We map what actually happens now, including the undocumented steps people do out of habit. That usually changes the scope.
The smallest build that works
Scoped to remove one specific task properly rather than to build a platform you didn’t ask for.
Failure handling
What happens when the other system is down or the data arrives malformed. Automation that fails silently is worse than no automation.
Handover
You get documentation of what it does and how to stop it, so you’re not dependent on us to turn it off.
Related work
Where this shows up in the portfolio
Questions
Straight answers
How do I know if something is worth automating?
Rough test: how many hours a month does it take, how settled is the process, and what does it cost when it goes wrong? A settled, frequent, error-prone task is worth automating. An occasional one that changes shape every month usually isn’t.
Is this AI?
Often not, and that’s fine. Plenty of the most useful automation is ordinary software doing something reliably. Where a language model genuinely helps — reading messy free text, classifying enquiries — we’ll use one, and we’ll say why.