We build AI systems.Then we evidence them.
Most of the work is the assurance around the systems, not the models. Data protection impact assessments, audit trails that survive being checked, and the evidence an inspector or an auditor will actually accept.
Two sectors, and we are building both out
Fixed price, fixed scope, and a document at the end that somebody else can check. The work in each is different. The way we do it is not.
AI is already in the service and the policy does not cover it
Social workers and foster carers are using consumer AI on their own devices. The authority is the data controller either way, so the liability sits with you whether or not anyone has written it down.
Somebody asked who can see your data and you could not answer
It comes with a customer security questionnaire, a Cyber Essentials renewal or an ISO deadline attached. We find out who can actually reach what across your Microsoft 365, and hand you the evidence to send back.
Tim Jackson and Dr Geoff Richards
Impending Tech works on the assurance around AI rather than on the models themselves. Data protection impact assessments, audit trails that hold up when somebody checks them, and the evidence a regulator or an inspector will accept. The people who sell you this are the people who do it.
We also build Chrysalis AI.
Some work cannot go near a public AI service. Case notes about children. Legal files. Anything where you would have to explain later exactly where the information went.
Chrysalis AI is a private AI appliance for that work. Hardware dedicated to one organisation, in the UK, and every action written to an audit log they can show an inspector. Same principle as the rest of this page: AI where it helps, and evidence that it did what you say it did.
Tell us what you already know. We will tell you whether we would find anything.
Either sector, or neither. If the answer is that you do not need us yet, that is a fine answer and it costs you one email.