What we do with your data, in plain language.

Client data is processed only for the engagement it was provided for, and only by the people working on it. Access is granted per project and removed when the engagement ends.

Credentials and secrets are held in environment configuration, never in source control, never in documentation, and never in anything we publish.

At the end of an engagement we return or delete client data on request, and confirm in writing which was done.

Where we process personal data on your behalf we act as a processor under your instructions, under a written agreement that records the scope, the retention period and the sub-processors involved.

We design for data minimisation: a system that does not need a field should not receive it. Where a pipeline can work on pseudonymised or aggregated data, that is what we build.

Subject access, rectification and erasure requests need a route through any AI system that stores derived data. We build that route rather than leaving it to be solved later.

Regulation (EU) 2024/1689 assigns obligations by risk tier. The first question on any engagement inside its scope is which tier the system falls into, because that determines what is required of it and when.

The Digital Omnibus, Regulation (EU) 2026/1744, in force 27 July 2026, deferred the high-risk obligations from 2 August 2026 to 2 December 2027 and 2 August 2028. We plan against the current timeline, not the original one.

Where a system is high-risk we build the technical documentation, logging and human-oversight provisions as part of the engagement, not as a compliance exercise afterwards.

Every system we ship has a written answer to three questions: what it will do, what it will not attempt, and what happens when it is uncertain.

We instrument for the failure mode specific to this class of system, which is confident wrongness. A wrong answer looks exactly like a right one, so it is tracked as its own event rather than lumped in with server errors.

We tell you when AI is the wrong tool. A simpler rule, or no change at all, is sometimes the honest recommendation, and we would rather say so than build something that should not exist.