One write-up per app. What it actually is, how we have used it in real work, what did not work, and where it stops being the right answer. Written from use, not from product pages.
Apps & AI Tools is the overview — the whole toolkit at a glance. This is the longer form. Each note covers one app in the detail a business owner actually needs to make a decision: what it is for, what it costs you in effort rather than licences, and the honest limits.
We write these after we have used something on real work, which is why the list grows slowly and why some entries say a tool did not stay.
Prototype. This section is a first pass for review. Brand marks and screenshots are being sourced; where an asset is not yet in place you will see a reserved frame rather than the finished thing.
Written up in full, from our own use.
01 · AI models & assistants
The model family, the shift from chat assistant to agent platform, and the difference between Claude Code, Cowork, the API and the Agent SDK — plus what we built on it, what broke, and why automated checks mattered more than prompting.
02 · AI models & assistants
We asked whether it could be our shared memory, because the files were chaotic. The answer was no — and the reason why is more useful than the tool. Where it landed instead: a rendered view, never operating memory.
Four tools we connected, configured and stopped using — a publishing pipeline that published nothing, a systems board that froze, local models that took the machine down, and a database we designed in and never needed.
Drafted or queued. Each one waits until we have something worth saying beyond what the vendor already says.
AI models & assistants
Where a second model earns its place, and the specific jobs we point it at rather than Claude.
Generative & creative
Running a second AI lane in parallel for image and media generation — a work-order protocol, and what it is honestly not good at yet.
Generative & creative
Connected as an integration and then blocked on re-authorisation — a note on connector lifecycle, not on output we have produced.
Dev, infra & web
Where our own sites and gated dashboards run, and why the boring infrastructure choice was the right one.
Knowledge & productivity
Using it as a visible governance layer rather than a second source of truth — and why that distinction matters.
Dev, infra & web
Why a non-software business ended up keeping its canonical records in version control.
These are in regular or occasional use and appear on Apps & AI Tools. They earn a note when we have something useful to add.
The point is never the tool. It is whether the business can see clearly and decide well — and which few tools, wired properly, make that true.
The first conversation is free and helps both sides decide whether the work is a fit.
Request a Consultation