Legal
Subprocessors
- Effective date
- Effective July 4, 2026
- Last updated
- Last updated September 24, 2026
This page lists every third party Agnotiq, Inc. engages to help operate Agnotiq MarginTide Pricing Intelligence (a subprocessor) on our behalf, and it is the authoritative, current list referenced by our Data Processing Addendum and Privacy Policy. We keep it current and commit to advance notice before we add or replace a subprocessor (§3).
Overview
A subprocessor is a third-party service provider we use to deliver part of the Service — for example, hosting our database, sending email, or performing AI inference. We require every subprocessor to protect personal information and Customer Data and to use it only to provide their service to us, under terms consistent with our Data Processing Addendum.
Current subprocessors
Each row below shows the vendor, what we use it for, the categories of data it may process, where it processes that data, and its engagement status. Active vendors are engaged for every workspace today. Beta — opt-in vendors are engaged only for workspaces that explicitly opt into a beta AI-inference provider or beta research provider — no workspace uses them by default.
| Vendor | Purpose | Data categories | Location | Status |
|---|---|---|---|---|
| Supabase | Managed Postgres database, authentication, and realtime; hosts account data and Customer Data, isolated per workspace using row-level security. | Account data, Customer Data, Authentication events | United States | Active |
| Anthropic | AI model inference for the research pipeline (Composer, Researchers, and the quality-grading step), unless a workspace supplies its own Anthropic API key (BYO Key). | Customer Data (research-pipeline inputs), AI-generated Output | United States | Active |
| Vercel AI Gateway | Routing and logging of AI model-inference calls for the structured-extraction worker (identity/llm_extract jobs), including provider fallback and per-call usage/cost accounting; team-wide zero-data-retention is off, so the gateway retains call logs per its own retention window. | Customer Data (extraction prompts), AI-generated Output, Request/usage metadata | United States | Active |
| OpenAI | AI model inference (via Vercel AI Gateway) for the structured-extraction worker, unless a workspace supplies its own OpenAI API key (BYO Key). | Customer Data (extraction prompts), AI-generated Output | United States | Active |
| Google (inference) | AI model inference (via Vercel AI Gateway) for the structured-extraction worker, unless a workspace supplies its own Google API key (BYO Key). | Customer Data (extraction prompts), AI-generated Output | United States | Active |
| Stripe | Payment processing and subscription billing, including validating and applying any promotion or coupon code redeemed; collects and stores payment details. | Billing data, Payment details | United States | Active |
| Slack | Delivery of alerts to a customer's Slack workspace, for workspaces that connect Slack via OAuth. | Workspace/channel identifiers, Alert content, OAuth credential | United States | Active |
| Resend | Delivery of transactional, notification, and onboarding/nurture emails, including magic-link sign-in. | Email address, Email content and delivery metadata | United States | Active |
| Vercel | Application hosting and content delivery for the Service, plus first-party web analytics, performance monitoring, and bot-abuse detection on form submissions (spec 1431). | Request metadata, Application logs, Anonymized page URLs and aggregate web-analytics/performance metrics (cookieless, no persistent identifier), Bot-detection signals | United States | Active |
| Sentry | Error monitoring and application diagnostics; error reports may include account identifiers and request metadata, never payment card data. | Account identifiers, Request metadata, Error diagnostics | United States | Active |
| Tavily | Research web-search — a web-search provider used by per-retailer research steps to look up competitor prices. | Search queries (derived from workspace configuration), Public web page content | United States | Active |
| ScrapingBee | Research page-extraction — a web-page retrieval and extraction provider used by per-retailer research steps to read competitor product pages. | Product/retailer URLs (derived from workspace configuration), Public web page content | France (vendor HQ); data processed in the European Union or United States per the vendor's disclosure | Active |
| Bright Data | Research page-extraction — a web-unlocking proxy/extraction provider used to fetch competitor product pages that block direct researcher fetches (`brightdata:extract:*` catalog rows). | Product/retailer URLs (derived from workspace configuration), Public web page content | United States (vendor infrastructure region per its own disclosure) | Beta — opt-in |
| Oxylabs | Research page-extraction — a structured e-commerce scraper API used to fetch and parse competitor product pages (`oxylabs:extract:parsed_*` catalog rows). | Product/retailer URLs (derived from workspace configuration), Public web page content | United States (vendor infrastructure region per its own disclosure) | Beta — opt-in |
| Zyte | Research page-extraction — a structured product-extraction API used to fetch and parse competitor product pages (`zyte:extract:product` catalog row). | Product/retailer URLs (derived from workspace configuration), Public web page content | United States (vendor infrastructure region per its own disclosure) | Beta — opt-in |
| Nexar | Procurement search — the Octopart electronic-component search API, licensed, used to look up distributor part offers for Procurement Intelligence workspaces (`nexar:search:octopart` catalog row). | Part-number search queries (derived from workspace configuration), Distributor offer data | United States (vendor infrastructure region per its own disclosure) | Beta — opt-in |
| Google (Places) | Services search — the Places Text Search API, licensed, used to look up service-provider listings for Services/SaaS pricing workspaces (`google:search:places_enterprise` catalog row). | Search queries (derived from workspace configuration), Place listing data | United States (vendor infrastructure region per its own disclosure) | Beta — opt-in |
| DataForSEO | Research web-search — Google Shopping product listings for price-checker research and public live-SERP page-1 paid-ad copy for marketing research (`dataforseo:search:google_merchant`, `dataforseo:search:google_serp_paid_p1` catalog rows). | Search queries (derived from workspace configuration), Public product listing and ad-copy data | United States (vendor infrastructure region per its own disclosure) | Beta — opt-in |
| Meta Platforms, Inc. | Marketing/ad search — the Ad Library API for competitor ad transparency lookups: looks up a workspace's watched competitors' running ads (`meta-ad-library:search:ads_archive` catalog row). | Public ad-archive metadata, Watched competitor domains (derived from workspace configuration); no customer PII | United States | Active |
Change notifications
Advance notice commitment.
Before we add a new subprocessor or replace an existing one, we will give at least 30 days’ advance notice. We provide that notice two ways: (a) an email to every workspace owner, and (b) an email to everyone enrolled in the notification signup below. You may object to a new subprocessor during the notice period by emailing legal@agnotiq.com; our Data Processing Addendum describes how we handle an objection.
Get notified.
Enter your email to be notified whenever this list changes. This is a compliance-notice list only — not a marketing subscription — and submitting the same address twice will never create a duplicate entry.
Questions
Questions about a specific subprocessor, our data-processing terms, or how to object to a change should go to Agnotiq, Inc. at legal@agnotiq.com. See also our Data Processing Addendum, Privacy Policy, and Security page.