Security
How Wendy keeps your data safe.
This page says where your data lives, who can reach it, and what we do if something goes wrong. Every sentence on it is true today. Nothing here is a plan.
Last updated October 8, 2026Previous versions →
Where it runs.
Wendy runs on Vercel in the United States (US East, iad1). Your data sits in Google Firebase, in Firestore and Cloud Storage, in the United States (nam5). Both platforms encrypt data in transit and at rest on their side; we do not run our own servers or disks.
What is never stored.
Message text. When you connect an account, a message is read into memory for one request, turned into a one-line note, and discarded. Learned from your communications, no messages are ever kept. Our logs carry ids, timing and status codes, never a message, an answer, a name or an email address; a redaction step strips those before a line is written.
How you sign in.
Sign-in is Google, Microsoft, or a six-digit code we email you. There is no password, so there is no password to leak. Under Profile & settings you can sign out everywhere, which ends every session on every device. Deleting your account does the same at once.
Who can reach your data.
You. Your data lives under your own account, and only our server writes it: the database rules refuse every write from a browser or an app and every read of another person's data. No Wendy employee reads your answers, your report, your chats, your notes or your memory. Our staff tools show counts, statuses and ids, and a staff decision that removes, restricts or hands over an account or its data — a minor review, a rights request, a deletion, a budget change — is written to an audit log kept for twelve months.
A workspace sees each member's profile and nothing else: no answers, no conversations, no notes, no inbox, no usage. What it can and cannot see is written out on its own page.
Connected accounts.
Each connection is your own grant, made by you, with read-only access, and you can end it under Profile & settings. We never ask a company for access to its people's accounts. The access token is encrypted before it is stored, with a key that lives only in our server's configuration, and each token is sealed to its own record, so a token copied out of one record cannot be opened from another. Keys can be rotated without a break. When you disconnect, we revoke the grant with the provider and delete the token.
AI processing.
Wendy runs on Claude models from Anthropic. Nothing we send is used to train models, by us or by them; our contract with them says so. Each call carries a one-way hash of your account id, never your id or your email. How long Anthropic keeps what we send depends on which model does the work, and the table on our AI page is kept current.
Payments.
Stripe takes and holds your card. Your card number never touches our servers; we keep a Stripe customer id and the status of your plan.
Deleting and taking your data.
Delete my account, under Profile & settings, ends every session now, and the deletion is carried out in full within thirty days. Download my data gives you everything as a zip: your answers, your report, your card, your notes, the people and interactions Wendy holds, and your conversations.
If something goes wrong.
If we find a breach, we contain it at once: we revoke keys, rotate the token encryption key, and turn off the affected part of the service. Within a day we work out whose data was involved from our audit rows, which hold ids and data classes, never content. We tell affected workspace customers within forty-eight hours of confirming what happened. Where the law requires it, we tell the supervisory authority within seventy-two hours. We tell affected people without undue delay when the risk to them is high, and we post a notice on this page when a breach touched more than a handful of accounts or any workspace. Afterwards we write up what happened and what we changed.
Report a security problem.
Write to support@askwendy.ai. A person reads it and replies within two business days. We do not threaten legal action against anyone who reports a problem in good faith.
Certifications.
We hold no security certification and do not claim one. If we ever do, it will be named here with its date.
Who we are.
Wendy is AskWendy LLC, a Virginia limited liability company, at 4726 16th Rd N, Arlington, VA 22207. A procurement question, or anything else about this page, goes to support@askwendy.ai.
Version.
This page: version 1, October 8, 2026.