Software that fixes how the work moves.

Most projects arrive the same way: someone has a business that works, a process that does not, and three quotes that all say something different. The first useful thing we do is usually not writing code — it is working out which of the three quotes is describing the actual problem.

After that it is fairly ordinary. We scope it in writing, build it in weekly increments you can see, and hand it over with the accounts in your name. The projects on this site range from a gelato brand's storefront to a secure legal documents system for a US law firm, which sounds like a lack of focus until you notice they are all the same job: take something a business does manually and make software do it reliably.

The AI work grew out of the same place. Nobody who called us wanted “AI”. They wanted lead triage that did not depend on someone checking an inbox, or property valuations that did not take an afternoon. That is what Lynx.ai turned into, and it is still how we scope anything with a model in it: find the manual step first, decide whether a model actually beats a rule second.

We work with clients in South Africa, the Netherlands, and the UK, entirely remotely. If you want to know whether we are any good, the case studies are more honest than anything we could write here.