Client stories: the question each business couldn't answer, what we built and what changed.
All workTools we built, for clients and for ourselves, when nothing we could buy did the job.
AI moves fast. Here's how we use it in our own work, and where to start in a business your size.
The AI hub
We use AI every day, and we try it on our own work first. Here's what we believe, what we won't do, and why we think it's all harder to keep up with than it should be.
Since we wrote this: nothing's changed yet. When something does, we'll add it to our list of changes, with the date.
There are too many tools and too much noise.
Every vendor says they're "AI-powered" now, even the ones that weren't last year. Most of the time it's an old feature with a new label, and working out which tools are actually useful takes more time than most people have.
You're stuck between two worries.
You're worried about falling behind, and you're worried about wasting money on tools that don't pay off. Both are fair.
Even we find it a lot.
We use these tools every day and we still find it hard to keep up. So if you're finding it hard, it's not just you.
Ben and Cam
We haven't got it all figured out either. But we've tested what works on our own business first, and there are a few things we won't do.
It's how we cover tracking, builds, automations and checks that would otherwise take more people.
What you get out depends on the rules you give it, the context it has and what it's connected to. The same model does poor work with no setup and good work with a clear one, so we spend more time on the setup than on choosing a model.
AI writes the first draft. A person decides what happens to it, and owns the result.
Put AI on top of bad tracking and bad data and you get confident mistakes. Get your measurement right first, then add the AI.
A person reads everything before it goes live.
We tell you which tools touch your data.
If we can't point to the numbers, we don't say it.
A hook blocks Claude Code's email, chat and publishing tools unless a person asks in that moment. Our bots write only to our own chat, CRM and pull requests.
The safeguards behind each one are on our how we work page, and every change we make is dated on our updates page.
See how we workWe try everything on our own work first. Every opinion we publish has a date on it and gets reviewed on a schedule, and when we change our mind, we say so publicly.
Nothing with that status yet.
The AI we trust most to change our code. We run it inside Cursor, with our own rules and a person reading every pull request.
What would change our mind: Another coding agent that sticks to our rules and review habits better.
Last known verdict
The best place we've found to run Claude Code: it's an editor, it syncs with GitHub and our plan covers plenty of use.
What would change our mind: A better place to run Claude Code, or a plan change that makes Cursor a poor fit.
Last known verdict
If you'd like a straight answer on where a business your size should start, we're happy to give you one.
