What that looks like
Some of what we’ve built, from two-day jobs to systems running for years:- Giving project managers their own cost and revenue numbers, live, out of an accounting system that had no way to share them safely.
- Posting supplier invoices straight into the ERP, so nobody keys them through a portal first.
- Working out which project a machine was on from where it actually was, rather than from a form filled in afterwards.
- Reading every note from a week of field inspections, and emailing management what mattered.
- Getting machine telemetry and GPS into the system crews already have open, instead of a portal only head office logs into.
- Texting someone the moment their timesheet looks wrong, while they can still remember the shift.
- Building a sales pipeline dashboard the CRM wanted a plan upgrade for, without standing up a database to do it.
- Letting payroll and project costing read the same timesheet, instead of asking staff to file two.
- Putting QR tags on physical assets so the code still resolves after the systems behind it change.
How we work
“Simplicity — the art of maximizing the amount of work not done — is essential.” — Principle 10, Agile Manifesto
Right-sized, not one-size
Sometimes the answer is a process change, a spreadsheet, or not automating at all. And
keeping card numbers and personal details outside what we build means the rest doesn’t need
fort-knox security.
The best software isn’t always the best fit
Best-in-class doesn’t help if the pricing breaks as you add people, or your own data sits
behind an upgrade. We push what you already run to its limits before buying anything new.
No two setups are the same
We arrive with no preferred stack and no fixed way of working. Some clients have an in-house
tech team, some a data office, some neither. We’ll work with yours, bring our own developers,
or use a shared setup we run ourselves.
Organised data pays for itself
Getting data out of the systems holding it and into one real database is where the leverage
is. It turns a report someone has to remember to open into an alert that finds them — and
it’s what anything later builds on, AI included.
How we use AI
We build with AI every day. Given the choice between organised data with no AI, and AI on top of a mess, we’d take the organised data every time.AI for the exceptions, not the norm
The standard path should be plain deterministic code. AI earns its place on the cases that
don’t fit the pattern — cheaper to run, and more predictable.
Tokens to build, not to run
We’ll spend tokens writing software; we’d rather not spend them every time it runs. Building
this way is cheap enough that small automations clear the bar — and once written, they’re
ordinary code.
Clients
We built one ourselves
QRtub
Most QR tools bake a URL into a sticker and hope nothing changes. We needed one that didn’t, so
we built it. If you’re rolling out QR codes, use it — it works with the systems you already run
rather than replacing them.