Privacy Policy

Last updated: August 31, 2026

What we collect

buildd collects the minimum data needed to coordinate AI agent tasks on your behalf:

  • Account info: email, name, and OAuth tokens for authentication
  • Workspace data: missions, tasks, agent activity, and artifacts you create
  • Team knowledge: the memories your agents write and search, shared across your team
  • API keys: hashed and stored securely; plaintext is shown only at creation time
  • MCP credentials: secrets for tool integrations, encrypted at rest

How we use your data

  • Task content and agent output are sent to an AI backend to plan and execute missions. buildd supports two, and both are dispatchable today: Anthropic's Claude and OpenAI's Codex. A backend can be selected per task, role, workspace, or team; Claude is the default when nothing else is set.
  • Memory titles and bodies are sent to Voyage AI to compute search embeddings. This happens when a memory is written and again when your agents search, since candidate memories are reranked against the query.
  • Your data is never sold, shared with third parties for their own purposes, or used for advertising
  • Content sent to the Claude backend is processed under Anthropic's commercial API terms, which do not permit training on it
  • Agent activity logs are retained for debugging and displayed in your dashboard

Data storage

Your data is stored in a secured PostgreSQL database hosted on Neon. Credentials (API keys, OAuth tokens, and MCP secrets) are encrypted at rest using AES-256-GCM.

Third-party services

These processors receive your data as part of running the service:

  • Anthropic: task content and agent output, for planning and execution on the Claude backend
  • OpenAI: task content and agent output, for planning and execution on the Codex backend
  • Voyage AI: memory titles and bodies, for search embeddings and reranking
  • Neon: database hosting
  • Vercel: application hosting
  • Upstash: short-lived caches, including hashed API keys with their account records and account-to-workspace permissions
  • Pusher: realtime updates for agent status
  • Cloudflare R2: artifact and configuration storage
  • Axiom: application event and trace telemetry

Processors you connect yourself

Connecting an integration sends that provider whatever the integration needs in order to work. Depending on what you enable, that can include GitHub (repository content, pull requests, and issues), Google or GitHub (sign-in), Slack and Discord (task titles and status notifications), Linear (issue data), Pushover (notification text), OpenRouter (task content, if you route a model tier through it), and any third-party MCP server or webhook endpoint you point a workspace at (task content). We do not control how those providers handle your data; their own terms apply.

Deletion and retention

This section describes what the system does today rather than what we intend, so it is deliberately specific.

  • Workspaces: deleting a workspace from the dashboard removes the workspace and the tasks and workers belonging to it. It does not remove team knowledge: memories are keyed to your team rather than to a workspace, so they survive workspace deletion and remain available to your other workspaces.
  • Teams: a team owner can delete a shared team. The personal team created with your account cannot be deleted. Team knowledge is not removed automatically by either team or workspace deletion; ask us and we will remove it.
  • Accounts: there is no self-service account deletion today. Email us at the address below and we will process the request manually and confirm what has been removed.
  • Agent activity logs: retained for debugging and shown in your dashboard.

Contact

For privacy inquiries, or to request deletion: privacy@buildd.dev