Nine dimensions roll into one Zoe Score. On the example deal it reads 64 out of 100.
Explore the Deal PackagePrivacy
Plain-English principles plus the full policy. Last updated October 27, 2026. Questions go to privacy@zoediagnostics.ai.
These four terms carry the same meaning everywhere in this document. Read means one paid diligence run on one company, and the signed document you receive at the end of it. Company means the business a Read examines. Company contact means the person at the company who grants Zoe access to that company’s systems. Zoe Score means the one composite number that grades a company’s health across the nine dimensions.
Zoe collects three kinds of data for a Read. From connected systems: the records needed to compute the Read, such as who communicates with whom, how often, in what cadence, and across which channels. In Slack the analysis is on patterns, not paragraphs: message text is scored and then dropped, message bodies are never stored, and personal identifiers are stripped at ingestion. From the investor: the company’s pitch deck, any data-room PDFs or CSVs, and the company contact call transcript, which are read in full to build and test the Claim Ledger. From the company: its acceptance of the data processing agreement, logged as a consent event, and the OAuth grants it makes. Zoe only reads through those grants, though the permission a provider grants can be broader than what Zoe reads, and Zoe never holds the company’s credentials. We also collect basic account information (name, email, company) when you sign up, technical telemetry for product analytics, and business visitor information from our public website. Business visitor information can include page and referral data, IP address, device information, company, and professional contact details matched by our visitor identification provider. These website systems are separate from the data behind a Read.
The data behind a Read is used only to build the Claim Ledger, check its claims against connected systems, generate and score company call questions, compute the Zoe Score with its confidence percentage and coverage tier, answer Ask Zoe questions from that Read, and produce the signed document you receive at the end of it. Account information is used to operate the service and communicate with you. Technical telemetry is used to improve the product. Business visitor information is used to understand which companies are exploring Zoe and to support relevant business outreach. We do not sell or rent personal data. We do send website usage data to advertising platforms and a business visitor identification provider, which several privacy laws treat as sharing for advertising purposes. The sections below name every platform and every processor and say what each one receives.
All data is encrypted in transit and at rest. Each customer’s Reads sit in that customer’s workspace, and sensitive fields are also encrypted with AES-256-GCM. Zoe stores synced records and the evidence behind a Read. Deleting a workspace deletes its data from our active systems and revokes its grants, except the audit log, the record of who accepted our terms and this policy, and any promo-code redemption or payment receipt, which are kept. Ask Zoe questions and answers already copied to Zep are not deleted with the workspace. The signed document is deleted with the workspace. See our /security page for the per-connector data handling table.
The honest answer is a list, not a word. Zoe does not sell or rent personal data. Data does go to four kinds of recipient. Advertising platforms: Google Ads, Meta, LinkedIn and Reddit receive page addresses, referrer, IP address, device information and conversion events from our website, and Google and Meta also receive them from our application. Measurement vendors: Google Analytics, Microsoft Clarity and PostHog receive page and interaction data, and Clarity and PostHog record replayable sessions of visits to our website. Visitor identification: RB2B, a Retention.com service, receives public website visit data and may match a United States visitor to business identity, company and professional contact information. Processors acting on our instructions under contract: hosting, database, payments, demo booking, CRM, email delivery, model inference and connector authorisation. The four sections that follow name every one of them and the data each receives. The data behind a Read, meaning the deck, the data-room files, the company contact call transcript and what is read from connected systems, goes only to processors and is never sent to an advertising platform, analytics vendor or RB2B. We disclose data otherwise only when required by law or to protect against fraud or imminent harm. One exception by design: Zoe presents its data processing agreement to the company that granted access and logs acceptance before any system is connected.
Three vendors measure how people use our website and our application. Google Analytics 4 (measurement ID G-H56BZ0Q21L) records page views, page addresses, referrer, device, browser, language, screen, approximate location derived from your IP address, and named events for viewing a key page, submitting the contact form, booking a demo, starting a sign-up and completing a sign-up. It sets the first-party cookies _ga and _ga_H56BZ0Q21L on .zoediagnostics.ai, and it sends data to Google’s advertising domain, so Google can join it to advertising identifiers. Microsoft Clarity (project vo4yvwtqde) records replayable sessions of visits to zoediagnostics.ai: mouse movement, clicks, scrolling, rage and dead clicks, keystroke timing, and the rendered content of the pages you saw, together with your IP address, user agent, referrer and page address. Clarity sets _clck and _clsk and, per its own project configuration, Microsoft advertising identifiers, which means Clarity data is joinable to Microsoft’s advertising graph. Clarity masks the content of our sign-up checkout and our connection pages. PostHog (United States, us.i.posthog.com) records page views and page leaves, automatically captures every click and form interaction including the text of what was clicked, captures web vitals, dead clicks and rage clicks, registers first-touch attribution on every event, and records replayable sessions of visits to zoediagnostics.ai, with what you type into form fields masked and our application and our connection pages excluded from recording. Our sign-up page is recorded, but your email address and Stripe’s payment form are kept out of the recording. If your browser sends a Global Privacy Control signal, PostHog makes no session recording. PostHog sets an identifier in a cookie and in local storage on .zoediagnostics.ai, so it is also sent to app.zoediagnostics.ai and api.zoediagnostics.ai, and it retains session recordings for 30 days. Clarity and PostHog are session recorders, not page counters, and this policy previously did not say so.
Our website and our application load a Google Tag Manager container (GTM-TQF7LGB4), which loads four advertising vendors. On our website, when they load depends on where you are. In the European Economic Area, the United Kingdom, Switzerland, Brazil, Turkey and Quebec, none of them loads until you accept advertising in the banner shown on your first visit. Everywhere else they load with the page, and you can switch them off from the link at the foot of every page, titled Do not sell or share my personal information in the United States; if your browser sends a Global Privacy Control signal, we treat it as that opt-out and they do not load. Our application still loads the container when a page loads, with no consent step. Google Ads (conversion ID AW-18096487023) receives the page address, referrer, IP address, user agent and the advertising click identifier, records conversions for contact, demo booked and sign up, and builds remarketing audiences through Google’s doubleclick.net advertising domain; its conversion linker writes first-party _gcl cookies. Meta (pixel 923323890519854) receives a page view on every page of our website plus contact, lead, schedule and registration events, sets the _fbp identifier and reads the _fbc advertising click identifier; on our application Meta receives the visit and the hostname but not a page view event. The LinkedIn Insight Tag (partner 8976156) receives the page address, referrer, IP address and user agent on every page of our website, records two conversion events, and allows LinkedIn to match a visit to a LinkedIn member; it also loads a tracking image that works when JavaScript is disabled. The Reddit pixel (a2_iqlugmccsovj) records a page visit on every page of our website and sets the _rdt_uuid identifier. Two limits we can state and you can verify in the container: Google Ads enhanced conversions are switched off and Meta advanced matching is switched off, so no hashed email address or telephone number is sent to either from our tags. Two facts that cut the other way: restricted data processing is off for Google Ads and limited data use is off for Meta, so both operate in full personalisation mode. Because the container also runs inside our application, Google receives the addresses of application pages; since 17 September 2026 the container strips the access-token portion of invitation, share, survey and follow-up links before that address is sent, so the link secret itself is not transmitted.
Our public website uses RB2B, a Retention.com service, to identify some United States business visitors. RB2B receives the page address, referrer, IP address, browser and device information, and similar identifiers from a visit. It may match those signals to a company and professional contact record. When a match is available, RB2B sends the matched business identity, company information, professional contact information and visit context to Zoe. Zoe writes that information to Twenty, our CRM, where it is clearly marked as originating from RB2B. We use it to understand account interest and support business outreach. RB2B does not receive a company’s pitch deck, data-room files, connected-system data, Zoe Score or any other data behind a Read. This identification is separate from signing up, submitting a form or connecting a source, and it runs only where advertising and visitor identification is allowed, as Cookies and tracking below describes.
These vendors process data on our behalf under contract rather than for their own advertising. Netlify hosts both zoediagnostics.ai and app.zoediagnostics.ai and therefore processes every request, including IP address, user agent and page address; Netlify Forms also receives and independently retains a copy of every form you submit, including your name, work email, company or firm, role, any free-text message or notes, and the campaign parameters, referrer and landing page attached to the submission. Google Cloud hosts the Zoe API, so it processes every API request and the application logs. Supabase is the product database and holds workspaces, accounts, connected-source records, Reads and the audit log; on our Slack connection pages your browser also contacts Supabase directly. Stripe processes payments: card details are entered into Stripe’s own frame on our sign-up checkout and never reach our servers, and Stripe sets the __stripe_mid and __stripe_sid identifiers and collects device signals for fraud detection, which it uses across every site it serves. Cal.com powers demo booking and receives your name, email address, time zone and your answers to the booking questions, together with the campaign parameters and the PostHog identifier we attach to the booking. Twenty is our CRM: every form submission and every demo booking is written there as a contact record plus a note that reproduces every field you typed, verbatim, including free-text messages, together with the campaign parameters, referrer and landing page of your first and latest visit, and when you buy, the purchase is added to that record with the plan, the amount, the date and Stripe’s payment reference, never card details. Where analytics is allowed, as Cookies and tracking below describes, our website also keeps a count of your earlier visits in your browser’s local storage, such as how many visits and days, and how often you viewed our pricing, demo, sign-up, sample and comparison pages or finished a tool; a form you submit or a demo you book carries a one-word level and a short list of signal names from that count, so Netlify, Cal.com and Twenty receive them, and the count itself stays in your browser. HeyReach sends our LinkedIn outreach messages and Smartlead sends our outreach email, and both receive the name, professional contact details, company and message history of the people we contact; Smartlead passes each send, open, link click, reply, bounce and unsubscribe on to Twenty. A link in one of those messages carries a reference to your record in Twenty: our website removes it from the page address before any analytics, advertising or identification tag can read it, keeps it in local storage for up to 120 days where analytics is allowed, and attaches it to a form you submit or a demo you book, so that when you reply, book or write to us we can update that record and pause the rest of the outreach. Resend delivers our transactional email and receives the recipient address, the subject and the full message body. OpenRouter is the gateway for every model call and every embedding in the product, so it receives deck text, data-room document text, connected-source excerpts and Ask Zoe questions and answers, and it routes them onward to Anthropic for analysis and to OpenAI for embeddings. Zep stores Ask Zoe conversation memory and receives the full text of each question and answer. Nango and Composio broker the authorisations to a customer’s own systems and hold the resulting access tokens, and Nango also proxies the reads, so it sees the records returned. Google provides sign-in for the application, so Google learns that an account signed in to Zoe and when. Google Fonts serves the typefaces on app.zoediagnostics.ai and on one legacy page of our website, which sends your IP address, user agent and referring page to Google on those loads; it sets no cookie. The customer’s own connected systems, Slack, HubSpot, Salesforce, GitHub, Atlassian Jira and Intuit QuickBooks, receive only the authorisation request, our read queries and, when a connection is disconnected, a request to revoke its access; data flows from them to us on the customer’s grant. Where a customer configures an outbound webhook, we post event payloads to the endpoint that customer chooses.
You have the right to access, correct, export, or delete your personal data at any time. To exercise these rights, email privacy@zoediagnostics.ai. We respond within 30 days. We honor these rights for every user, regardless of jurisdiction.
We use cookies, tracking scripts and similar identifiers for four purposes: keeping you signed in, product analytics and session replay, advertising measurement, and business visitor identification. Our sites do set third-party advertising cookies and load cross-site tracking pixels. Google Ads writes first-party _gcl advertising cookies through its conversion linker and sets test_cookie on doubleclick.net; Meta sets _fbp and reads _fbc; Reddit sets _rdt_uuid and _rdt_pn; Microsoft Clarity sets _clck and _clsk alongside Microsoft advertising identifiers; and the LinkedIn Insight Tag sets cookies in LinkedIn’s own domain. Google Analytics sets _ga and _ga_H56BZ0Q21L, PostHog sets its own identifier in a cookie and in local storage, Stripe sets __stripe_mid and __stripe_sid for fraud detection on our checkout, and RB2B uses a tracking script and similar identifiers to match some United States business visitors. Because most of these are written on .zoediagnostics.ai, they are sent to our application and our API as well as to our website, even where the vendor’s script does not run there. On zoediagnostics.ai, what loads depends on where you are. In the European Economic Area, the United Kingdom, Switzerland, Brazil, Turkey and Quebec, a banner asks first, separately for analytics and for advertising and visitor identification, and none of these loads until you accept; declining sets nothing except a record of your choice. Everywhere else they load with the page, and the link at the foot of every page lets you switch either purpose off. In the United States that link is titled Do not sell or share my personal information, and a Global Privacy Control signal from your browser switches advertising and visitor identification off automatically. Switching something off removes the identifiers it set on our domain and stops it loading. Stripe’s cookies on our checkout are used for fraud prevention and load regardless. Our application, app.zoediagnostics.ai, does not yet ask: the tag manager container loads there when a page loads. You can block or delete cookies in your browser settings and use the opt-outs each vendor publishes; the site will continue to work with the exception of features that require login. If you want your data removed from these vendors rather than merely blocked going forward, email privacy@zoediagnostics.ai and we will action it under the rights section above.
Zoe operates from the United States. If you access the service from outside the US, your data will be transferred to and processed in the US under standard contractual clauses or equivalent safeguards. EU customers can request that the data behind a Read be processed in EU regions.
We may update this policy as Zoe evolves. Material changes will be communicated to active customers via email at least 30 days before they take effect. The "last updated" date above reflects the most recent revision. The September 2026 revision corrected two statements that were not accurate: that Zoe shared data with nobody, and that our sites used no third-party advertising cookies or cross-site tracking pixels. Both are now described as they actually operate.
Questions about this privacy policy?
privacy@zoediagnostics.aiInspect the controls
Review the access model, evidence trail, and boundaries before you invite a company.