Case study
Our own private AI platform
We run our own AI workflows on open-weight models on our own hardware, with budgets, logs, approval steps and tests.
What we built
Our research, document reading and content checks all run on these models, under the same controls.
The controls on every workflow
- Every workflow calls the models through a gateway with its own key, a daily budget and a kill switch.
- Anything from outside is screened for hidden instructions before a model sees it.
- Actions that matter wait for approval and are logged in a tamper-evident record.
- Prompts are checked against a held-back test set.
Where the AI runs
Tier 3
A private model on hardware in the UK
Open-weight models on your hardware or ours. Nothing goes to an outside AI provider. Our own research and document reading run this way.
Storage location and processing location aren’t the same thing. We’ll tell you both, in writing.
Related pages
Get in touch
Want something like this? Tell us the job.
One paragraph is enough. We’ll reply with honest thoughts on whether we can help, and what it would cost.