# Conduit — the AI connector gateway

Conduit gives you 3,000+ connectors for Claude, Cursor, ChatGPT, or any other AI assistant, with the controls and visibility your IT team needs to say yes.

## How it works

- **Connect** — point any MCP client at a single endpoint, users sign in through the identity provider with SSO.
- **Access** — 3,000+ prebuilt connectors out of the box, plus any MCP server, OpenAPI definition, or GraphQL endpoint you bring.
- **Control** — groups and policies decide who can use which connectors and tools. Every call is authenticated, authorized, and logged.

## Supported AI clients

- Claude
- ChatGPT
- Cursor
- VS Code
- Gemini
- Windsurf
- Any MCP-compatible client

## For employees

- **Your tools are already there** — 3,000+ connectors with prebuilt tools, ready to enable. No waiting for someone to build the integration.
- **Sign in once** — authenticate with your company account. Conduit handles the OAuth dance with every downstream tool for you.
- **Your credentials stay yours** — the model calls tools through Conduit. It never sees your tokens, passwords, or API keys.
- **Keep the assistants you have** — conduit is a standard MCP server. If your AI tool speaks MCP, it works — no plugins, no migration.

## For IT & security

- **Policies, not tickets** — policies grant connectors to the entire workspace, select groups, or individual users, down to the individual tool. Access changes take effect immediately.
- **Your identity provider** — sSO through Okta or any OIDC provider. Provisioning rules automatically route users into groups and roles from IdP claims at sign-in.
- **Credential isolation by design** — end-user credentials are held by the gateway and never exposed to the LLM or the client.
- **Audit everything** — every admin change and every tool call lands in the audit log, scoped per workspace.
- **Observability you already run** — openTelemetry traces and events export to your existing stack — Datadog, Grafana, anything OTel.
- **Deploy it your way** — self-host a single Docker image in your own infrastructure, or use the managed cloud.

## Links

- [Pipedream Connect](https://pipedream.com/connect)
- [App marketplace](https://pipedream.com/apps)
- [Pricing](https://pipedream.com/pricing)
- [Contact us](https://pipedream.com/contact)