How Ask Nathan uses your data — and what it never does

Ask Nathan works from what your team connects (HubSpot or Salesforce, which it reads and never writes to; Google and Microsoft calendars, for meeting history only; Microsoft 365 email, for who wrote to whom and when, only for a member who switches it on; and Google Sheets, read-only) and from what it adds: LinkedIn export files, attendee and target lists, spreadsheets and notes, the documents you add about what you sell, and the meetings and confirmations your team logs. We do not scrape LinkedIn. We do not buy contact databases. We do not call third-party APIs to enrich without you knowing.

Company names you save as your team’s target list are stored with your workspace and visible to its members. Company names you check on the target list page without saving them are still sent to our servers to check coverage.

Where you choose to share relationship metadata — for example, that a teammate has met a buyer — Ask Nathan stores the signal, not the underlying message content. Owners control what is shared and can remove signals at any time.

When your own meetings or your company’s CRM show a colleague connected to a company you’re asking about, Ask Nathan may suggest inviting that colleague. It says only that they were in a meeting with you or own the deal in the CRM, never more than that fact. A colleague can ask not to be suggested, through the link in an invite email or by contacting us.

Every warm path, intro ask, and outreach draft is labeled as a draft and lists what to verify before sending. Ask Nathan sends a message in a team member’s name only when that member chooses to send it: a Verified Intro email reaches the person you want to meet, in your teammate’s name via Ask Nathan, only after the teammate who knows them chooses to forward it. If that person accepts, Ask Nathan sends one email, in the same teammate’s name, that introduces you. Ask Nathan does not claim relationships your team hasn’t confirmed.

The public Coverage Gap tool (asknathan.ai/coverage) follows the same rules. A visitor who connects Google or Microsoft Calendar there isn’t a signed-in customer — Ask Nathan reads only meeting times and attendee addresses from the last 90 days, never the subject, description, or a contacts list, and only to compute two numbers: how many people at that company you can already reach, and how many more through someone you both know. That calendar data is used once, to render that one result, and is discarded before the response is even built — never written to a database, a log, or an analytics event.

Ask Nathan only reads from HubSpot and Salesforce and never writes to them, Google and Microsoft calendars connect for meeting history only, and Google Sheets connect read-only, once someone on your team connects them. Salesforce’s API permission has no read-only option, and the HubSpot app Ask Nathan connects through includes write permissions, so Ask Nathan asks for that access and only ever reads. Microsoft 365 email connects only for a member who switches it on, and reads who wrote to whom and when, never subjects, message text or attachments. Gmail and Slack are not part of this beta. Every connection is made by a member of your workspace, shows what it reads before it syncs, and can be revoked at any time.

Ask Nathan reads up to 3 years of meeting metadata from a connected Google or Microsoft calendar: who met whom, when and how often. It reads the most recent twelve months first, then the older months in stages, a portion with each sync, until all of them are read. After that, each sync refreshes only the most recent twelve months. Event titles, descriptions and contents are never read. Microsoft 365 email, for a member who switches it on, is read as far back as the calendar, most recent first: who wrote to whom and when, never subjects, message text or attachments.

These are the service providers that process customer data for Ask Nathan, and what each one handles. We update this list before adding a new one.

• Lovable Cloud (on Supabase): hosts the application and its database, so it stores all workspace data and sign-in details.
• Composio: holds the authorizations for the CRM (HubSpot or Salesforce), calendars, Microsoft 365 email and Google Sheets your team connects, and relays the data Ask Nathan syncs from them.
• Anthropic: writes the plain-language Why behind a path and intro drafts, from role placeholders, account names and evidence counts; reads an Ask-bar question when Ask Nathan can’t fully read it on its own (the whole question is sent); and suggests a company’s industry, type, size and region, and checks whether a company or a job title fits what you search for, from the company’s name, website and the job titles of people there and the words you search for, after removing contact details and the names of people in your workspace (a name Ask Nathan doesn’t hold can remain).
• Lovable Email: sends invite, intro and pilot-terms acceptance emails, so it handles the addresses they go to.
• PostHog (US): product analytics, as usage events with a pseudonymous id and your company’s domain.
• Stripe: processes payments and subscriptions for plans bought online. It holds the buyer’s billing name, email, address, tax ID and payment method, and receives the ids of the buyer’s workspace and account so a purchase can be matched to it. It receives no other workspace data.

Every Path Confidence score is explainable. Open any path and see the Why: the evidence behind it, as kinds, counts and dates. The score is computed the same way every time, and a model never sets it. There are no “black box” scores in the product.

Ask Nathan is built by Nathan Pettyjohn, founder/president of the VR/AR Association and the Applied AI Association, and former Lenovo XR GTM lead. The team is hiring its founding technical co-founder and founding engineers.