PAI Sub-processors
PAI is operated by Techzoom Ltd (company no. 15539411). We use the third parties below to help run PAI. Each is bound by a data processing agreement and processes personal data only on our instructions. Where a sub-processor is outside the UK, transfers are covered by the UK IDTA or the UK Addendum to the SCCs, plus a transfer risk assessment.
Last updated: 6 July 2026. We will update this list before adding or replacing a sub-processor. If you would like to be notified of changes, email privacy@pai.archi and we will let you know before they take effect.
| Sub-processor | Purpose | Personal data processed | Location | Transfer safeguard | DPA |
|---|---|---|---|---|---|
| Vercel | Web application hosting and server functions | Data passing through the application, including conversation content in transit; request logs | Global edge network; server functions in London, UK (lhr1) | UK IDTA / Addendum (US-headquartered) | DPA |
| Clerk | Authentication and identity | Name, email, user ID, sign-in metadata | USA | UK IDTA / Addendum | DPA |
| Anthropic | AI chat assistant (Claude) | Conversation content; uploaded-file content sent as prompts | USA | UK IDTA / Addendum | DPA |
| OpenAI | Embeddings for semantic search | User search queries; public case-document text | USA | UK IDTA / Addendum | DPA |
| Cloudflare (R2) | File storage and database backups, encrypted at rest | Uploaded files and extracted text; database backup copies | Western Europe (WEUR) | UK IDTA / Addendum (US-headquartered) | DPA |
| Stripe (from paid launch) | Payments and billing | Billing name and email, plan and transaction history; card details are collected and held by Stripe, not us | USA, with local processing entities | Stripe DPA (incl. UK Addendum) | DPA |
| Twilio SendGrid | Transactional email | Email address, name, email content | USA | UK IDTA / Addendum | DPA |
| Sentry | Error monitoring | Error diagnostics, which may include identifiers and request data | EU | UK IDTA / Addendum (US-headquartered) | DPA |
Notes:
- Anthropic and OpenAI are accessed through their commercial APIs. Under those terms, customer inputs and outputs are not used to train their models.
- Our backend, database, cache, task queue and semantic search reranker all run on our own dedicated servers in the UK, so stored data involves no third-party hosting sub-processor.
- Some providers hold data in the UK or EU but are US-headquartered. We keep UK transfer safeguards in place with them as well, since a US parent company can be subject to US law.