How we handle your data.
We advise clients on data governance, so publishing nothing about our own would be an odd position to hold. This is current practice, stated plainly.
We take the least data that will work
Assessments run on samples and schemas wherever a sample will answer the question. Production data comes into scope when the build needs it, not at the discovery stage.
Your data stays in your tenancy where it can
Preference is always to build inside your cloud account, your database, and your identity provider. Where that is not possible we agree the exception in writing before anything moves.
Model providers are named before we use them
You will know which provider processes your data, under which terms, and whether that provider trains on it. Enterprise API terms that exclude training are the default, and we will tell you when they are not available.
Access is per person and time-boxed
Named individuals, scoped to the systems the engagement needs, revoked at the end of it. No shared credentials and no standing access after handover.
Return or destruction on request
At the end of an engagement, or at any point you ask, working copies of your data are returned or destroyed and we confirm in writing which happened.
Mutual NDA before the detail
Available before the first conversation. You do not have to describe your problem to a stranger without one.
We measure our own advertising. This site sets a cookie of its own recording which campaign or link brought you here, and Meta's advertising pixel runs on every page. When you send a form or book a review, that record goes to Meta together with a one way hash of your email address, which Meta matches against accounts it already holds. We do not send the address itself, and nothing you write in a message or a booking note is sent in any form. Blocking third party scripts in your browser stops the Meta half, and everything here still works.
We advertise in ChatGPT, and measure that too. OpenAI's advertising pixel runs on every page here and reports the page view. When you book a review, that booking is also reported to OpenAI from our server, with a one way hash of your email address so they can match it against accounts they already hold. We do not send the address itself, we send no name at all, and nothing you wrote in a booking note is sent in any form. Nothing else on this site is reported to them: a signup or a message through the contact form is not. Blocking third party scripts stops the browser half, and everything here still works.
We keep a log of our own traffic. Every page view here writes a row in our own database: which page, when, which link or advertisement sent you, your IP address, and what your browser reports about itself, which is its user agent, its window and screen size, its languages and its timezone. We also record how long the page stayed open and how far down it was read. It is kept for six months and then deleted, it stays on our server, and none of it is sold, shared or sent to an advertising network. Nothing in it carries your name unless you have given it to us on a form.
We work out roughly where you are. This site sends your IP address to a geolocation service, ipwho.is, and keeps the country, city and timezone it answers with. It is how we know which markets our advertising is reaching, and what time it is where you are before anybody telephones you. Your address is all they are sent. Your name, your email and anything you wrote are not. The answer is remembered for a month, so one visitor is looked up once however many pages they read. Treat it as an estimate rather than a location: it is often a city out, and behind a VPN it is a country out.
Questions about any of this, or a security questionnaire that needs completing, go to a person here.
Bring us the pilot that stalled.
Bring the security questionnaire too, if there is one.
You'll leave with 2 to 3 scored use cases, an effort estimate, and an honest cost range, whether or not we work together.