Security and trust
Your company’s knowledge stays under your control, and the AI tools can only read it.
What IT and procurement teams usually ask: who controls your knowledge, where your data goes, how long we keep drafts, and where it’s all processed.
01 · Control
BeigeCheck proposes, and people approve.
Your brand guidance and approved facts change only when a person approves it. BeigeCheck keeps its suggestions apart from those decisions.
Approved company knowledge
Your documents and writing become proposed facts and voice guidance. An owner or approver checks each one before any AI tool gets it.
People decide
A person approves every change to your company’s knowledge, and the person writing decides whether a draft goes out. The checks inform that decision.
A record of every change
Each approved change is kept as a version, with who approved it and when, so you can see how your guidance changed and go back.
02 · Where information goes
Your knowledge, the drafts you check, and what your AI tool sees.
Step 1
Your source material
Brand guidelines, documents and writing go into your company’s workspace, or are read from folders you connect.
Step 2
Approval
An owner or approver reviews what BeigeCheck proposes from them before it’s used anywhere.
Step 3
Your AI tool
When someone asks for BeigeCheck, their AI tool reads the approved knowledge it needs, signed in as that person. The connection can only read.
Step 4
Checking
A draft is checked for phrasing that reads as AI-written and for claims your approved knowledge doesn’t back.
Worth knowing: a claim your approved knowledge doesn’t back may still be true. BeigeCheck shows where the evidence is missing, so you can add it or leave the claim out.
03 · Drafts
You choose what happens to the drafts you check.
An owner chooses on the Team page how long your company keeps draft text. Whichever you choose, we keep the scores, and a shorter choice clears drafts already kept for longer.
Never keep draft text
We keep the scores and none of the draft’s words.
Keep drafts for 90 days
Then a daily job clears the words.
Keep drafts for 365 days
Then a daily job clears the words.
04 · Where data lives
Your data is processed in the UK and Europe, apart from sign-in and the AI model.
| What | Provider | Where |
|---|---|---|
| Database | Supabase | Ireland |
| Servers | Vercel | London |
| Sign-in | WorkOS | United States |
| AI model for checks | Anthropic | United States |
| Mailgun | European Union |
- Encryption
- Data is encrypted in transit and at rest. Keys for connected folders are stored encrypted, and only our servers use them.
- Separation between companies
- Our database keeps every company’s data apart from every other company’s with row-level security.
- AI model training
- We don’t use your data to train AI models, and our AI provider’s terms don’t allow it to train on what we send.
- Who can do what
- Owners manage the workspace and the team, approvers decide on changes, and writers use what’s approved. Each person signs in as themselves, and owners can see which AI tool each person has connected.
- Transfers outside the UK and Europe
- Where data leaves the UK or Europe, we rely on the UK International Data Transfer Agreement or the EU Standard Contractual Clauses.
- Contracts
- For your company’s workspace, you’re the data controller and we’re your processor. We’ll sign a data processing agreement.
05 · Questions
Questions worth asking before you connect.
Will BeigeCheck train AI models on our data?
Who can change our approved knowledge?
Can other companies see our information?
Do you keep the text we check?
Does a check prove every claim is true?
Can our IT or procurement team review BeigeCheck?
The full detail is in our privacy policy and terms of use.
Get the answers you need before you start.
Whether you’re trying it with a small team or planning a wider roll-out, we’ll answer your questions plainly.