Legal
Sub-processors
Version: 1.0 · Effective date: 7 October 2026 · Last updated: 7 October 2026
About this list
This page lists the third parties that Smart Unlimited Holding B.V. (trading as AnyForge) ("AnyForge") engages to process Customer Data, as referred to in Section 7 and Annex III of our Data Processing Addendum and Section 7 of our Privacy Policy. It also lists, separately, the AI providers and other services that customers connect themselves, which are not AnyForge sub-processors.
Changes. Before a new sub-processor begins processing Customer Personal Data, we update this page and email the owners of every customer organization at least 30 days in advance, as described in Section 7.2 of the DPA. Customers may object on reasonable data protection grounds within that period. Questions or objections: privacy@anyforge.ai.
Sub-processors engaged by AnyForge
AnyForge engages these providers on its own account. They act on our instructions under written data protection terms.
| Provider | Purpose | Data | Location | In use |
|---|---|---|---|---|
| Google Cloud and Firebase (Google LLC) Data terms | Hosting and infrastructure: database (Firestore), file storage, application hosting, Cloud Functions and Cloud Run compute, task queues, secret storage, logging, analytics warehouse (BigQuery) and sign-in (Firebase Authentication). | All Customer Data stored in or processed by the Service; Account data (name, email address, sign-in records); Provider credentials (encrypted in Secret Manager); Logs and usage data | European Union: Netherlands (europe-west4) and the eur3 multi-region (Netherlands and Belgium); task queues in Belgium (europe-west1). Firebase Authentication is a global Google service. | Always |
| Google Cloud Vertex AI (Google LLC) Data terms | Text embeddings (model text-embedding-004) that power semantic search: AnyForge converts text into numeric vectors so the Oracle, crews and search can find relevant code, documentation and library documents. Runs on AnyForge's own Google Cloud account, not a customer key. Not used to train models. | Source code from connected repositories (in chunks); Codebase analysis documentation; Context Library documents, including meeting summaries and transcripts a user shares to it; Search queries typed in the Oracle and in search | Netherlands (europe-west4) | When a repository is analysed or indexed, a Context Library document is added, or a semantic search runs |
| Google Cloud Speech-to-Text (Google LLC) Data terms | Live transcription of Google Meet audio for the Meeting Assistant. The audio is not stored by AnyForge. | Meeting audio, including the voices of everyone in the call; The resulting transcript text | European Union (eu multi-region) | Only when an organization turns on the Meeting Assistant and a user starts audio capture |
| Google Analytics (Google LLC) Data terms | Website and product usage analytics (page views and visits). Not loaded on shared-link pages. | Pages visited; Device and browser information; IP address (used by Google to derive approximate location); Analytics cookie identifiers | United States and other Google locations | Always |
| PostHog Data terms | Product analytics: a small number of server-side product events (for example, that a crew was started). In-browser PostHog analytics is currently switched off. | User and company identifiers; Product event names and properties | European Union (PostHog EU cloud) | Always |
| Stripe Data terms | Subscription billing and payment processing for AnyForge plans. | Billing contact details; Company billing details; Payment card data (held by Stripe, never by AnyForge); Subscription and invoice records | United States and other Stripe locations | When an organization subscribes to a paid plan |
| Resend Data terms | Transactional email (invitations, notifications, digests) and receiving email sent to an organization's AnyForge intake address. | Recipient and sender email addresses and names; Email content, including the text and attachments of emails sent to an intake address | United States | Transactional email always; email intake only when an organization turns it on |
| WhatsApp Business Platform (Meta) Data terms | The WhatsApp channel to the Oracle: messages a user sends to AnyForge's WhatsApp number and the replies. | Phone number and WhatsApp profile name; Message content and media the user sends; Replies sent by the Oracle | Meta locations, including the United States and Ireland | Only when a user links WhatsApp and messages the AnyForge number |
| Discord Data terms | AnyForge's community server: optional membership linking for the community request board, and announcements of shipped bug reports and feature requests users submitted to AnyForge. | Discord user id and guild membership (when a user links Discord); Title and summary of feedback a user submitted to AnyForge | United States | Only when a user links Discord, or submits feedback to AnyForge that later ships |
| StackBlitz WebContainers Data terms | The in-browser terminal in Code Studio. The runtime is loaded from StackBlitz into the user's browser and package installs run through StackBlitz's infrastructure. | IP address and browser information; Names of packages installed from the terminal | United States | Only when a user opens the Code Studio terminal |
Providers you connect
These providers receive data only when your organization connects them with its own key, account or installation, and only the data your configuration routes to them. Under Section 8 of the DPA they are engaged by you, under your own agreement with them, and are not AnyForge sub-processors.
| Provider | Purpose | Data | Location | In use |
|---|---|---|---|---|
| Anthropic Data terms | Claude models for the Oracle, crews, Code Studio, triage and analysis, on the customer's own Anthropic API key or Claude subscription. | Prompts, code and content the customer routes to the model; Model outputs | As set in the customer's Anthropic account | When the customer connects an Anthropic key or subscription |
| OpenAI Data terms | OpenAI models, on the customer's own OpenAI API key. | Prompts, code and content the customer routes to the model; Model outputs | As set in the customer's OpenAI account | When the customer connects an OpenAI key |
| Google AI Studio (Gemini API) Data terms | Gemini models, on the customer's own Google AI Studio key. | Prompts, code and content the customer routes to the model; Model outputs | As set in the customer's Google account | When the customer connects a Google AI Studio key |
| Google Cloud Vertex AI (customer project) Data terms | Claude models served from the customer's own Google Cloud project, with the customer's service account. | Prompts, code and content the customer routes to the model; Model outputs | The region the customer chooses | When the customer connects a Vertex AI service account |
| Amazon Bedrock (Amazon Web Services) Data terms | Claude and other models served from the customer's own AWS account. | Prompts, code and content the customer routes to the model; Model outputs | The AWS region the customer chooses | When the customer connects AWS Bedrock credentials |
| OpenRouter Data terms | Access to models from many vendors through one API, on the customer's own OpenRouter key. | Prompts, code and content the customer routes to the model; Model outputs | United States; the serving model vendor depends on the model the customer picks | When the customer connects an OpenRouter key |
| Z.ai (Zhipu AI) Data terms | GLM models, on the customer's own Z.ai or BigModel key. | Prompts, code and content the customer routes to the model; Model outputs | As set by the endpoint the customer chooses (Z.ai international or BigModel, China) | When the customer connects a Z.ai key |
| DeepSeek Data terms | DeepSeek models, on the customer's own DeepSeek key. | Prompts, code and content the customer routes to the model; Model outputs | China | When the customer connects a DeepSeek key |
| Baseten Data terms | Models hosted in the customer's own Baseten account. | Prompts, code and content the customer routes to the model; Model outputs | As set in the customer's Baseten account | When the customer connects a Baseten key |
| TypeSafe Data terms | Independent second-opinion checks on selected crew decisions, on the customer's own TypeSafe key. | The crew decision being checked and its context; The verdict returned | As set in the customer's TypeSafe account | When the customer connects a TypeSafe key |
| Self-hosted model endpoints | A model server the customer runs itself (an OpenAI-compatible endpoint such as vLLM). | Prompts, code and content the customer routes to the model; Model outputs | Wherever the customer hosts it | When the customer registers its own endpoint |
| GitHub Data terms | Reading and writing the customer's repositories, pull requests and reviews, through the AnyForge GitHub App installed by the customer and its operators' own GitHub sign-in. | Source code and repository metadata; Pull requests, reviews and comments; GitHub usernames | As set in the customer's GitHub account | When the customer installs the GitHub App or an operator connects GitHub |
| Slack Data terms | The Oracle, triage and notifications in the customer's Slack workspace. | Messages and files in channels and threads where AnyForge is used; Slack user names and ids | As set in the customer's Slack workspace | When the customer installs the AnyForge Slack app |
| Microsoft Teams Data terms | Approval and halt notifications posted to a Teams channel through an incoming webhook the customer creates. | Notification text: crew, work item and approval names and status | As set in the customer's Microsoft 365 tenant | When the customer adds a Teams webhook |
| Sign-in identity providers | Signing in with Google, Microsoft, GitHub or the customer's own SAML or OIDC provider. | Name, email address and the identifier the provider shares | As set by the identity provider | When a user signs in with that provider |
| Atlassian (Jira and Confluence) Data terms | Reading and updating the customer's Jira issues and Confluence pages through the Atlassian connector. | Issues, pages and comments the connector reads or writes | As set in the customer's Atlassian site | When the customer connects Atlassian |
| Linear Data terms | Syncing work items with the customer's Linear workspace. | Issues, projects and comments that are synced | As set in the customer's Linear workspace | When the customer connects Linear |
| Zendesk Data terms | Help desk intake: filing tagged tickets as requests and adding completion notes. | Ticket content and requester name and email | As set in the customer's Zendesk account | When the customer connects Zendesk intake |
| Intercom Data terms | Help desk intake: conversations the customer's Intercom workspace sends to AnyForge. | Conversation content and requester name and email | As set in the customer's Intercom workspace | When the customer connects Intercom intake |
| Datadog Data terms | Reading monitors, logs and traces for investigations, and receiving alerts. | Alerts, logs, traces and metrics the customer's Datadog account returns | As set in the customer's Datadog site | When the customer connects Datadog |
| Sentry Data terms | Receiving error alerts the customer's Sentry account sends to AnyForge. | Error titles, stack locations and alert metadata | As set in the customer's Sentry account | When the customer points a Sentry alert at AnyForge |
| Aikido Security Data terms | Reading the customer's security findings. | Security findings for the customer's repositories | As set in the customer's Aikido account | When the customer adds an Aikido token |
| Vanta Data terms | Reading the customer's compliance tests and evidence. | Compliance controls, tests and evidence status | As set in the customer's Vanta account | When the customer adds a Vanta credential |
| Amazon Web Services (customer account) Data terms | Reading cost data and fetching private packages (CodeArtifact) from the customer's own AWS account. | Cost and usage figures; Package names and versions | The AWS region the customer chooses | When the customer adds AWS credentials |
| Stripe (customer account) Data terms | Reading revenue figures from the customer's own Stripe account for business-value reporting. | Revenue and subscription figures | As set in the customer's Stripe account | When the customer adds a Stripe key |
| Snowflake Data terms | Read-only queries against the customer's Snowflake warehouse. | Query results the customer's warehouse returns | As set in the customer's Snowflake account | When the customer adds Snowflake credentials |
| Databricks Data terms | Read-only queries against the customer's Databricks workspace. | Query results the customer's workspace returns | As set in the customer's Databricks workspace | When the customer adds Databricks credentials |
| Temporal Cloud Data terms | Reading workflow status from the customer's Temporal Cloud namespace. | Workflow names, status and history | As set in the customer's Temporal Cloud account | When the customer adds Temporal credentials |
| Harness Data terms | Reading pipelines and deployments from the customer's Harness account. | Pipeline, deployment and service metadata | As set in the customer's Harness account | When the customer adds a Harness token |
| Source-code escrow agent (SFTP) | Delivering source-code escrow deposits to the SFTP server of the escrow agent the customer appoints. | Source code and build documentation in the deposit | Wherever the escrow agent hosts it | When the customer configures escrow deliveries |
Your organization can also connect other tools through MCP servers it adds itself. Data goes to those servers only as your configuration directs.