What happens to your data.
Chuck works on occurrence reports, investigation findings and your manuals. This is where they go, and what we commit to.
What we commit to.
Your data doesn’t train AI models
Not ours, and not our model provider’s. It’s in the contract, so there’s no setting to find and switch off.
Zero data retention at the model provider
Your question and the passages sent with it exist at the provider only while the request runs. Nothing is kept afterwards.
Your documents stay yours
What you upload is stored on our own infrastructure in the UK, and only your organisation can read it, unless you send us an answer to review.
Kept apart from other organisations
Another organisation can’t open your files, chats, projects or library, and every route is tested for it.
Cite or refuse
Every regulatory claim shows the passage it came from. When the regulations don’t support an answer, it says so.
Advisory, never the decision
Chuck proposes and cites. A named person in your organisation decides, and every document it produces has a line for them to sign.
Checkable before it’s used
Every grounded answer shows its passage. Citations and quotes are checked by software that isn’t AI, and every document says to check it before use.
What it reads is material, not instructions
Chuck treats what it reads in your documents as material, not instructions, and tells you when a document seems to be addressing it. It can still be misled, so the safeguards that matter are built in: it has no way to send your data to other sites, and links to sites other than ours and the regulators’ ask before they open.
Where it runs: three places, set out separately.
- Data at restUnited Kingdom · European Union
- Your documents, the regulation library and your account are stored on our own infrastructure in the UK. Backups are encrypted here before they go to Amsterdam, in the EU.
- Inference, at our model providerEuropean Union
- Your question and the passages used to answer it go to our model provider, encrypted, for the length of the request. It stays in the EU, and nothing is kept or used for training.
- Processing on our own serversUnited Kingdom
- Indexing your documents, reading the text out of your files and dictation run on our own UK infrastructure, with no outside provider. Dictation audio is never stored.
The sub-processors page lists every sub-processor and its region, and the privacy notice covers personal data.
The documents behind it.
Security & Trust Overview
The full overview, for your security and procurement teams. We send it during evaluation and go through it against your questionnaire.
Request the overviewData Processing Agreement
Sent with the overview during evaluation, for the terms that govern your data.
Request it with the overviewSub-processors Register
Every sub-processor and its region.
Read the registerPrivacy Policy
How we handle personal data.
Read the policy
Common questions
Does our data train AI models?
No. Not ours, and not our model provider’s. It’s in the contract, so there’s no setting to find and switch off.
Where is our data stored and processed?
Your documents, the regulation library and your account are stored on our own infrastructure in the UK, and backups are encrypted here before they go to Amsterdam, in the EU. Your question and the passages used to answer it go to our model provider in the EU for the length of the request. Indexing your documents, reading the text out of your files and dictation run on our own UK infrastructure, with no outside provider.
What does the model provider keep?
Nothing. Your question and the passages sent with it exist at the provider only while the request runs, and nothing is used for training.
Who can read what we upload?
Your organisation, unless you send us an answer to review. Another organisation can’t open your files, chats, projects or library, and every route is tested for it.
Why not use a general-purpose assistant?
A general model has no curated regulations to draw on. It can cite a rule that doesn’t exist, and it can’t show where an answer came from. Business versions of the major assistants come with real commitments; the risk is a report pasted into a personal account, where inputs may be used for training.
Can an approved organisation use it?
Your organisation stays accountable for everything it does with a tool like this, and Chuck is built for that. It proposes and cites, a named person in your organisation decides, and any answer downloads as a Word record of the question, its sources and the checks, with a blank section for your review.
The full Security & Trust Overview, on request.
During evaluation we send it to your security and procurement teams, with the Data Processing Agreement, and go through it against your questionnaire.