2.1 Account and sign-in information
You can sign in to WHIPT with Google, with GitHub, or with a one-time code we email to you. Sign-in is handled by our authentication provider, WorkOS. When you sign in we receive and store:
- your name, as provided by your sign-in method (if none is provided, we use your email address as your display name);
- your email address and whether it has been verified;
- a unique account identifier issued by WorkOS;
- the organization (workspace) memberships and roles that link your account to a WHIPT workspace.
During sign-in, your IP address and browser user agent are passed to WorkOS for security, fraud prevention and rate limiting. We also keep short-lived records of sign-in attempts (keyed to your email address and IP address) to block repeated failed attempts.
2.2 Google user data (Sign in with Google)
This section describes exactly what we do with data we receive from Google when you choose Sign in with Google.
What we access. We request only the openid, email and profile scopes. From these we receive your Google account identifier, your email address, whether Google has verified that email address, and your basic profile information (your name and, if set, your profile picture URL). We do not request access to your Gmail, Google Drive, Google Calendar, contacts or any other Google data when you sign in.
How we use it. We use this data only to create and sign you in to your WHIPT account, to show your name and email address inside the product (for example, to your teammates in the same workspace), to send you service messages about your account, and to keep your account secure. We do not use it for advertising.
How we store it. Your name, email address, verification status and account identifier are stored in our production database hosted by DigitalOcean in the United States, and in your sign-in record at our authentication provider, WorkOS. Our database does not store your Google profile picture (WorkOS may keep the picture URL as part of your sign-in record). Your session is kept in an encrypted, secure browser cookie. We never receive or store your Google password.
How we share it. We share Google user data only with the service providers that help us run the sign-in and hosting of the Service (WorkOS for authentication, DigitalOcean for hosting and database, Cloudflare for network delivery), as described in Section 5. Our email delivery, hosted AI and error monitoring providers (Section 5.1) do not receive your Google user data for sign-in or account management; they may see your name or email address only where it appears in Workspace Content, such as a task or message that mentions you. We do not sell Google user data, we do not transfer it to data brokers or advertising platforms, and we do not use it to train generalized artificial intelligence or machine learning models.
How to delete it. You can ask us to delete your account and the Google user data associated with it at any time by emailing privacy@whipt.ai from the email address on your account. See Section 10. You can also remove WHIPT's access to your Google account at any time at https://myaccount.google.com/permissions.
Limited Use. WHIPT's use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements.
Google Workspace connectors. Separately from sign-in, a workspace administrator may choose to connect Google Workspace apps (such as Gmail, Google Drive or Google Calendar) so that WHIPT agents can work with that data. You see and approve the specific Google permissions at the time you connect. Data we receive through those connections is used only to perform the tasks you or your workspace direct, is subject to the same Limited Use commitments above, is not used for advertising, is not sold, is not used to train generalized AI or machine learning models, and is not read by our staff except with your permission, where needed for security, or where required by law. You can disconnect a connected app inside WHIPT or at https://myaccount.google.com/permissions at any time.
2.3 Workspace content you provide
When you use WHIPT you and your teammates create and upload content, including company and project information, agent configurations and instructions, issues, comments, approvals, decisions, files and attachments, published pages and funnels, automations, and the outputs that AI agents produce ("Workspace Content"). Workspace Content is stored in our database and in private object storage (Cloudflare R2) and is used to provide the Service to your workspace.
Company memory. WHIPT agents can save and recall memories for your workspace (for example facts, decisions and summaries of past work). The text of each memory is stored in our database. Search by meaning is not yet turned on for customer workspaces (Section 5.2). When it is, to let agents find memories by meaning, we may also send memory text to an AI model provider through OpenRouter to create an embedding (a list of numbers that represents the meaning of the text), and store that embedding in Qdrant Cloud, our vector search provider. Qdrant stores only the embedding and identifiers such as the workspace, memory record and agent identifiers; it does not store the memory text. When a memory is deleted or corrected, or a workspace is deleted, see Section 7 for how the embeddings are removed.
2.4 Connected apps and integrations
If your workspace connects third-party apps (for example GitHub, Slack, Google Workspace or other tools in the WHIPT app directory), we store the credentials or access tokens needed for that connection in encrypted form, and we exchange data with that app as your workspace directs. Some connectors call third-party services that process the text your agents send them; where we know of this, WHIPT shows a notice before you connect.
2.5 Customer inbox and published pages
If your workspace enables a customer inbox, we process the emails sent to and from that inbox, including the names and email addresses of senders and recipients, subject lines, message bodies, email headers (such as message identifiers, reply references and sender authentication results) and attachments. We send and receive that email through an email delivery provider, AgentMail or Resend, depending on how the inbox is set up (see Section 5.1). If your workspace publishes pages or funnels with forms, we process what visitors submit through those forms. We do this as a processor for the workspace that owns the inbox or page. Published pages may also record cookieless visitor statistics (see Section 6).
2.6 Usage, log and device information
Our servers record technical information when you use the Service, such as the pages and API endpoints requested, timestamps, error details, browser type, and IP address. In our web server access logs, IP addresses are truncated before they are written: we keep only the first 24 bits of an IPv4 address (/24) and the first 48 bits of an IPv6 address (/48). Where we keep an IP address in a security or audit record (for example a request to join a workspace), we store it truncated in the same way. We use this information to operate, secure and troubleshoot the Service. On our checkout pages, Stripe and Cloudflare Turnstile also collect device and browser signals to prevent fraud and automated abuse (Section 2.11).
We send error reports to Sentry (see Section 5.1) so we can find and fix problems. Error reports contain the type of error, its message and the program locations where it happened (a stack trace), and, for errors on our servers, the request method, the API path requested and non-sensitive request headers (such as the browser user agent). Our software configures Sentry not to collect default personal information, and it does not attach your name, email address or account identifier to error reports. When your browser delivers an app error report, Sentry necessarily sees your browser's network connection, but our configuration does not ask Sentry to record your IP address. Browser error reporting runs only while you are signed in and stops when you sign out. Browser reports leave out the page address, query string, referrer and click, console and network histories. Server reports leave out query strings, request bodies, cookies, authentication and signature headers, and IP addresses, and replace database and network error text with a fixed message. Session replay (screen recording) is not used. An error message can occasionally contain a fragment of the content that caused it.
We use PostHog product analytics in the app (see Section 5.1) to understand how the Service is used, so we can improve it. App analytics runs only while you are signed in, and it stops and clears its identifier when you sign out. It records the pages you view and the type of each control you use, with page addresses reduced to their route (identifiers in the address are replaced with a placeholder, and the query string and fragment are removed). It also records the page you came from (reduced to its website address, or to its route if it is part of the app), any campaign tags in the link that brought you to the app (such as utm parameters and ad click identifiers), the search engine you came from and any search terms it passes along, your browser, device and screen information, and the approximate location (such as city, region, postal code and country) that PostHog derives from your IP address. PostHog receives and stores your IP address. It identifies you by your internal account identifier and your workspace by its internal identifier, never by your name or email address. Page titles are not sent, the text and attributes of the elements you click are masked, and session replay (screen recording) and surveys are not used. Our servers also send a small number of account events to PostHog (signing up, creating a workspace, inviting a member and joining a workspace). They contain internal identifiers and a few event details, such as the sign-up method, the member's role and the kind of workspace, never names, email addresses or content, and PostHog does not derive a location for them.
2.7 Feedback on AI outputs
If you rate an AI output (thumbs up or down), your vote is saved in your workspace. If you choose "Always allow" to share voted outputs, we may review the shared outputs and votes to improve the quality of WHIPT. You can change this choice at any time in Settings.
2.8 Communications
If you email us or contact support, we keep your message, your contact details and our reply so we can help you.
2.9 Payment information
Payments are processed by Stripe (see Section 5.1). When you buy credits, you enter your card details into fields or pages that Stripe hosts, either on our checkout pages at https://pay.whipt.ai or on a Stripe payment page. Your full card number and security code go directly to Stripe; we never receive or store them.
When you check out, we collect the name and email address you enter, and we create a Stripe customer record for your organization. From Stripe we receive and store, for each card on file: Stripe's identifiers for the customer and the card, the card brand, the last four digits, the expiry month and year, and the card fingerprint (a value Stripe derives from the card number that lets us recognize the same card again without knowing its number). We also store Stripe's identifiers for each payment, refund and dispute. Card fingerprints are used only for fraud prevention (Section 2.11); we keep them out of API responses, logs and error reports, and our staff do not see them.
2.10 Credits, usage and billing records
WHIPT bills usage from prepaid credits held in a wallet for each company (workspace). Today we record and price usage but do not yet deduct it from wallets (see Section 9 of our Terms of Service). To run your wallet we keep:
- a record of each billable unit of usage, such as a hosted AI model request, an embedding, or an email sent, with the time, the type and quantity of usage, the provider's cost where one applies, the credits charged, and the workspace and agent it was for;
- a ledger of every change to the wallet balance (purchases, usage, refunds, disputes and adjustments), with the amount, the resulting balance, the reason, who or what made the change, and the related Stripe identifier;
- each payment attempt, including the credits being paid for, the price at the time, and which card it was charged to; and
- for card disputes, the dispute's status, reason, amount, currency and response deadline.
2.11 Fraud prevention and account security
To protect the Service and our customers from payment fraud and abuse, we use the following signals:
- Bot protection. Our checkout pages use Cloudflare Turnstile, which evaluates your browser and network signals, including your IP address, to tell people from automated traffic. Our servers check the result with Cloudflare on each checkout attempt.
- Stripe fraud screening. Stripe uses your IP address and device and browser information collected by its script on the checkout pages to screen payments for fraud.
- Card fingerprints. We compare card fingerprints across accounts to detect a card that is reused by another organization or that belongs to a banned account.
- Email checks. We refuse disposable email domains at checkout, and if an account already exists for the email address you enter, we send a sign-in code to that address instead of continuing the purchase.
- Account bans. If our staff ban an account for fraud or abuse, we record that it was banned, when, and the reason. We do not complete purchases made with any card that a banned account has used.
- Rate limits. We limit repeated checkout attempts by IP address, and by each combination of email address and IP address. Limits based on your IP address are evaluated in memory (for IPv6, using its first 64 bits) and are not written to a database. We also limit how many unpaid purchases can be open for each combination of email address and IP address; that count is kept in memory and checked against the purchase records, which do not store your IP address.