Star Subprocessors
Last updated: August 1, 2026
A subprocessor is a company we use to deliver the hosted Ganju service that may handle personal data on our behalf. This page is the authoritative list. It’s part of our Privacy Policy and our Data Processing Agreement.
If you self-host Ganju, none of this applies to you — you choose your own providers, and the only ones you inherit are the ones you configure yourself.
How we change this list
Before a new subprocessor starts handling Customer Content, we will:
- Update this page and the date at the top of it.
- Email the Owner of every organization on a paid plan at least 30 days before the change takes effect.
- Give you a way to object. If you have a reasonable, documented data-protection objection we can’t resolve, you may terminate your subscription for the affected service and we’ll refund the unused portion of your current period.
Emergency replacements — a provider that fails or is terminated for cause — may happen faster. We’ll tell you as soon as we can and explain why.
To get these notices, email hello@ganju.ai and ask to be added to the subprocessor notification list. We’ll also post changes to the same page you’re reading now.
Infrastructure
These handle data for every customer. You cannot opt out of them and use the hosted service.
| Subprocessor | Purpose | Data it handles | Location |
|---|---|---|---|
| Cloudflare, Inc. | Application hosting, CDN, object storage (R2), queues, containers, transactional email routing | All traffic; uploaded files; queued jobs; outbound email | United States / global edge |
| Neon, Inc. (on AWS) | Managed Postgres with pgvector — our system of record | Accounts, workspaces, Customer Content, text chunks and embeddings, conversations, audit and error logs | AWS us-east-1, United States |
| Google LLC (Gemini API) | Generating embeddings for every resource and search query | Resource text and search queries | United States / global |
| Stripe, Inc. | Subscription billing and payment processing | Name, email, billing address, payment method, usage counts | United States / global |
Google appears in this tier deliberately. Embeddings run on our key for every customer on every plan, so resource text reaches Google whether or not you’ve configured anything. See AI models, embeddings, and your content.
Conditional — only if you enable them
These handle data only for organizations that switch on the matching feature. If you never connect them, they never see anything of yours.
| Subprocessor | Triggered by | Data it handles |
|---|---|---|
| Google LLC (shared model) | Running channel replies on Ganju’s shared model — the default, and the only option on Free | Conversation history, system prompt, tool definitions and results, retrieved chunks |
| Anthropic, PBC | Adding an Anthropic model key | The same turn content, on your own account |
| OpenAI, L.L.C. | Adding an OpenAI or OpenAI-compatible model key | The same turn content, on your own account |
| Tavily | Installing the web-search or web-extract tools | Your search queries and target URLs |
| Telegram, Slack, Meta (WhatsApp), Discord | Connecting a channel on that platform | Messages to and from your bot, participant identifiers |
| Google, Microsoft, Slack, Cal.com | Connecting an account so tools can act on it | Whatever each tool request sends and returns |
Any remote MCP server you connect via mcp-proxy | Installing that server | Whatever your tools send it |
Any HTTP endpoint you configure via http-endpoint | Installing that tool | Whatever your tools send it |
The last two rows are destinations you choose. We can’t vet them, we don’t control them, and they are not covered by our commitments — they’re listed so the picture is complete.
Identity providers
| Subprocessor | Purpose | Data it handles |
|---|---|---|
| Google LLC | Social sign-in | Your name, email, profile image |
| GitHub, Inc. | Social sign-in | Your name, email, profile image |
Questions
Write to hello@ganju.ai or call +57 312 4678519.
Ganju S.A.S. · Bogotá, D.C., Colombia