Subprocessors

Last Updated: November 13, 2026

These are the providers that process personal data on behalf of Fiftyone Group LLC when you use 51 Intel. This page supports our Privacy Policy and our Security & Compliance page. We update it when the list changes.

In use means the provider is part of normal operation. When enabled means our code supports the provider but it runs only when a setting or credential is switched on, so it may not be receiving data at a given time.
The location column states only what the provider’s own published documentation confirms, or where we host a service ourselves. “Not stated” means we do not make a statement about where that provider processes data.
Where you choose a Slack, Microsoft Teams or webhook address as a destination for your alerts, that address is your own channel and not our subprocessor. We deliver the alert content there, and that service’s own terms apply.

Named Subprocessors

Provider	What it does for 51 Intel	Data it receives	Where it processes data	Status
Supabase	Database and sign-in (authentication).	Your account and profile details, sign-in records, consent records, usage and access logs, Ask 51 questions and answers, personal briefings, alert settings and share link records.	Database and backups are kept in the region selected for our project (Oregon, United States). Supabase states that it also processes account, support, telemetry and content-delivery data outside the selected region.	In use
Render	Hosts the 51 Intel API and its scheduled processing.	Everything a reader sends to or receives from the API while it is processed: questions, profile fields, requests and their network addresses.	Oregon, United States (our service region). Render states that its primary processing operations take place in the United States.	In use
Vercel	Hosts the website. API requests from your browser pass through Vercel on their way to the API. Also measures page load performance.	Page requests and API requests and responses in transit, including question text, IP address and browser details. Performance measurements are collected without cookies.	Not stated.	In use
OpenAI	AI models: plans and embeds Ask 51 questions, writes answers when it is the selected answer model, searches the web for outside research, and drafts personal briefings.	Question text, up to six recent turns of the same conversation, your saved Ask 51 perspective, profile fields (job title, seniority, company name and type, topics, regions, goal text) when answers are personalised, the 51 Intel records retrieved for the answer, and for briefings up to five of your most recent questions.	Not stated. We do not select a regional endpoint.	In use
Anthropic	AI models: writes answers when it is the selected answer model, runs deep research and agent research, reads web pages and uses its web search tool for outside research.	Question text, recent turns of the same conversation, your saved Ask 51 perspective, profile fields (job title, seniority, company name and type, topics, regions, goal text) when answers are personalised, retrieved 51 Intel records and web page text.	Not stated. Anthropic offers a US-only processing option that 51 Intel does not currently set.	In use
Firecrawl	Web search and page reading for Ask 51 research.	Search queries, which can include the words of your question, and the addresses of pages it is asked to read.	United States. Firecrawl’s privacy policy states that its servers are located in the United States.	In use
Exa	Web search for Ask 51 research.	Search queries, which can include the words of your question.	Not stated.	In use
Parallel	Checks the key statements of an Ask 51 answer against public sources, and is an alternate web search for Ask 51 research. Also used for 51 Intel’s own company research, which involves no reader data.	For the check: the answer’s key statements, not your question or profile. For search: search queries, which can include the words of your question.	Not stated.	In use
Inngest	Schedules and runs background jobs, including personal briefings, alerts, digests and data refreshes.	Job events that carry reader account identifiers, and the results of jobs, which can include briefing content.	Not stated.	In use
Resend	Email delivery: alert emails, digests, scheduled briefs and sign-in emails.	Your email address and the content of the email.	Not stated.	In use
Stripe	Payment processing and billing for paid subscriptions.	Your email address, name, payment card details and subscription status. Card details are collected by Stripe directly; we do not store card numbers. Stripe’s scripts load on the checkout page.	Not stated.	In use
Beehiiv	Newsletter membership checks, and newsletter tier updates when a subscription changes.	Your email address.	Not stated.	In use
Cloudflare (Turnstile)	Security check on the sign-in code request.	Browser and network signals, including your IP address, collected by the Cloudflare script when you ask for a sign-in code.	Not stated.	When enabled
Google Analytics 4	Product analytics.	Page paths without query strings or fragments, share link tokens removed, interaction events, and device and browser details. Loaded only after you accept analytics in the cookie notice.	Not stated.	In use
PostHog	Product analytics for signed-in readers.	Your account identifier (not your email address), page paths without query strings or fragments, and the controls you click. The text of what you click is left out on pages that show your own words, such as Ask 51. Loaded only after you accept analytics in the cookie notice.	Not stated.	When enabled
Slack (51 Intel’s workspace)	Internal notifications to our team, including feedback you submit in 51 Intel and internal digests.	For feedback: your email address, rating, comment and the page path.	Not stated.	When enabled
Instantly	Sends invitations to company representatives to review their company’s profile. Concerns company contacts, not readers.	Contact name, work email address and company.	Not stated.	When enabled
Attio	Read-only lookups of 51 Intel’s customer relationship records for those invitations. Concerns company contacts, not readers.	A company website domain, returning the company record, its deal stages and its contacts.	Not stated.	When enabled

AI and Search Providers: Retention and Training

These statements follow each provider’s current published terms. Those terms can change, and the links are the documents we relied on.

OpenAI: data sent to the API is not used to train or improve OpenAI models unless the customer opts in, and abuse monitoring logs are kept for up to 30 days by default (OpenAI data controls).
Anthropic: Anthropic may not train models on customer content from its commercial services (Commercial Terms of Service), and API inputs and outputs are deleted from its backend within 30 days of receipt or generation by default (Anthropic Privacy Center).
Zero data retention is not in place with OpenAI or Anthropic. Both may keep request data for a limited period for abuse monitoring under their terms.
Exa and Parallel are search and checking providers, not model providers for 51 Intel. Their published standard terms allow them to use queries or inputs to improve their own models, unless an enterprise zero data retention arrangement applies (Exa privacy policy; Parallel customer terms). We do not state that either provider is barred from training on this data.

Sources for Location Statements

Supabase: Data Residency and Transfers FAQ
Render: Data Processing Addendum
Firecrawl: Privacy Policy

Questions

For questions about this list, or to request a Data Processing Agreement as an institutional subscriber, contact privacy@fiftyone.xyz.