How Connections Work
Connectors are how AI teammates reach the tools and data they need. A connector describes an app, API, or data source Type can use. A connection is a configured instance of that connector, usually backed by OAuth, an API key, or custom API settings.
Connections give Type access to tools and data. A connected integration can be personal, using one user’s app permissions, or organization-level, using shared company access. After an integration is connected, add it to the teammates that should be allowed to use it.
What connectors unlock
Section titled “What connectors unlock”Connectors can let a teammate:
- Read and write in tools such as Slack, GitHub, Linear, Notion, Google Drive, Salesforce, HubSpot, Commslayer, Intelligems, and support platforms.
- Query data warehouses such as PostgreSQL, BigQuery, Supabase, Databricks, and Snowflake.
- Use custom APIs for internal systems.
- Receive events from apps that can trigger tasks.
Connection types and teammate access
Section titled “Connection types and teammate access”| Concept | Meaning |
|---|---|
| Personal connection | A connected integration using one user’s account and app permissions. |
| Organization-level connection | A connected integration using shared workspace or company access. |
| Global availability | Connected integrations are available in Type for assignment. |
| Teammate access | The specific connected integrations an AI teammate is allowed to use. |
| Custom API | Any internal or external tool with an API that does not already have a packaged integration. |
Auth methods
Section titled “Auth methods”Different apps connect in different ways:
- OAuth for apps that support delegated login.
- API keys or bearer tokens for tools that expose credentials.
- Service account JSON for some organization-level Google connections. BigQuery, Google Ads, and Google Analytics support either Google OAuth or a service account key. Google Analytics also asks for the GA4 property ID to report against.
- No-auth tools when a provider does not require credentials.
- Custom API configuration when you want Type to call your own service.
- Hosted MCP connections for tools that expose their own Model Context Protocol server, such as PostHog, Commslayer, and Intelligems.
Type stores OAuth grants and encrypted API keys securely. Credentials should not be pasted into teammate instructions or thread messages.
For Google Drive & Docs, Type asks for access to specific files and folders instead of broad document access. After connecting the integration, open its details from Connections and choose the files teammates may read or update, plus the folders they may use as destinations for new Google Docs and Drive changes. If a teammate thinks more Drive access is needed to finish a task, it can show a Select files card so you can add more files or folders without reconnecting Google.
Browse and connect apps
Section titled “Browse and connect apps”From a teammate, open its connector settings to search the app catalog and add connected integrations to that teammate. Type supports a large, regularly updated integration catalog.
You can also ask a teammate in conversation to use or connect an app. A template prompt may suggest likely connections, but selecting the template does not add or authorize them. During setup, Type asks before adding an available workspace connection. If the connector is not configured yet, Type can guide you into the connection flow.
Connector setup guides
Section titled “Connector setup guides”Some connectors require credentials you create in the source system before opening Type:
- Connect BigQuery with a Google Cloud service account JSON key.
- Connect Shopify with a Shopify app from dev.shopify.com, its Client ID, Client Secret, and access scopes.
- Connect Meta Ads with a Meta system user access token and ads permissions.
Connection guidance in prompt templates
Section titled “Connection guidance in prompt templates”Prompt templates describe likely or needed connections so you can review them during setup. They do not enforce a selection or connect an app automatically.
| Template | Connection guidance |
|---|---|
| Data Analyst | Live reporting needs one data source, such as PostgreSQL, BigQuery, Supabase, Databricks, or Snowflake. |
| Product Engineer | Repository work needs one source code host, such as GitHub, GitLab, or Bitbucket. |
| Other role templates | The prompt suggests likely connections for the role; confirm only the ones needed for your workflow. |
Do not paste credentials into the setup conversation. Complete OAuth, API key, or other authorization in the connection flow.
Custom APIs
Section titled “Custom APIs”Use a custom API connection when Type does not yet have a first-party integration for your tool. Provide the authentication details and API documentation needed for the teammate to understand what it can call. For the full setup flow, see Connect a Custom API.
Connectors and skills together
Section titled “Connectors and skills together”Connectors answer “what can this teammate access?” Skills answer “how should this teammate do this repeatable workflow?” A strong teammate usually has both: the right connectors and a small set of clear skills.