These vendors may process Customer Data to deliver Rahoto. The same list is published in the legal sub-processor schedule that the Data Processing Agreement refers to.
If your workspace connects its own AI model, the tasks you route to it are sent to that model’s provider under your own agreement with them; that provider is your vendor, not a Rahoto sub-processor, and Rahoto’s pseudonymisation still applies before the request leaves. The same holds for the data sources you connect and the delivery destinations you configure (for example your own Slack workspace, storage bucket or webhook): they are your vendors, not Rahoto sub-processors.
| Vendor | What it does | Data it receives | Location |
|---|---|---|---|
| Hetzner Online GmbH | Application servers and database hosting | All Customer Data at rest in the database | EU (Helsinki, Finland) |
| Cloudflare, Inc. | File storage (R2) for report assets, exports, snapshots and database backups; asset delivery; DNS | Report assets, exports, snapshots and database backups | Western Europe (Cloudflare R2 location WEUR) |
| Resend | Account, verification and delivery-notification email | Recipient email address and message content | United States (sending region as set in the Resend contract) |
| Stripe | Subscription billing and payment processing | Billing contact, billing address and payment details | EU / US (SCCs) |
| Sign in with Google; Gemini API as a configured AI provider, also embeddings and image generation | Sign-in: name and email. Gemini: Your prompt and your data’s structure and summary statistics as text, not its rows, plus what a task works on (the report’s own text and figures, anything you point Ask Rahoto at, passages from text sources you analyse). Names, addresses, emails and phone, card, bank account and national ID numbers in that text are pseudonymised first. Images of report pages, and scans and audio you send for transcription or text extraction, are processed as submitted. Not used for training under the provider’s API terms. | EU / US (SCCs) | |
| Microsoft | Sign in with Microsoft | Name and email address | As set in the Microsoft contract |
| Anthropic | Claude API, a configured AI provider for report drafting and reasoning | Your prompt and your data’s structure and summary statistics as text, not its rows, plus what a task works on (the report’s own text and figures, anything you point Ask Rahoto at, passages from text sources you analyse). Names, addresses, emails and phone, card, bank account and national ID numbers in that text are pseudonymised first. Images of report pages, and scans and audio you send for transcription or text extraction, are processed as submitted. Not used for training under the provider’s API terms. | US (SCCs) |
| OpenAI | OpenAI API, a configured AI provider; also audio transcription, embeddings and image generation | Your prompt and your data’s structure and summary statistics as text, not its rows, plus what a task works on (the report’s own text and figures, anything you point Ask Rahoto at, passages from text sources you analyse). Names, addresses, emails and phone, card, bank account and national ID numbers in that text are pseudonymised first. Images of report pages, and scans and audio you send for transcription or text extraction, are processed as submitted. Not used for training under the provider’s API terms. | US (SCCs) |
| TypeSafe | Advisory check that cited sentences in an AI draft agree with their sources | Sentences from report drafts and the claims they cite | As set in the TypeSafe contract |
| Recraft | Draws vector illustrations when you ask Ask Rahoto for one. Not used when your workspace requires EU-only processing or its data classification forbids cloud AI. | It receives only a short abstract description of the artwork, never report data, figures, names or files; the description is scrubbed for personal details first. | Outside the EU |
AI model providers
Which configured provider handles a request depends on the task, the deployment's configuration, provider health and your organisation's policy. Whichever it is, what leaves Rahoto as text is your prompt and the structure and summary statistics of your data, not its rows, plus what a task works on: the report's own text and figures, anything you point Ask Rahoto at, and passages from text sources you analyse. Personal details in that text, such as names, addresses, emails and phone, card, bank account and national ID numbers, are pseudonymised. Images of report pages, and scans and audio you send for transcription or text extraction, are processed as submitted. See how Rahoto handles your data with AI.
An AI model endpoint that your organisation configures itself is your vendor, not a Rahoto sub-processor.
Our commitments
- At least 30 days' notice before a new sub-processor handles Customer Data.
- You may object to a new sub-processor; we will work with you on an alternative or you may end the contract.
- When a vendor is removed, we stop sending it data and request deletion under our agreement with it.
Get notified of changes
Email support@rahoto.com with the subject "Sub-processor notifications". Enterprise customers are notified through their designated security contact.