AI Connector Agent Beta
Fivetran AI Connector Agent provides a fast and flexible way to generate connectors for SaaS data sources that make their data available through REST APIs, including sources that Fivetran does not currently support. Powered by artificial intelligence (AI), the agent generates production-ready connectors in minutes, eliminating lengthy development cycles and the need to write, maintain, and deploy connector code.
Generating a connector with the AI Connector Agent takes just a few steps. Provide basic details about your data source, such as its name and publicly available API documentation. The agent analyzes the API, identifies supported endpoints, and gives you control over what to sync. A tailored connection setup experience allows you to configure authentication and connection settings with ease. Before you start the initial sync, you can review and approve the connection schema.
The AI Connector Agent is designed for users who want a Fivetran-managed experience for connector generation. Connectors generated by the agent are deployed on Fivetran infrastructure, so you can focus on selecting the data you want to sync instead of managing connector hosting or deployment. If your source does not provide data through REST APIs, or if you prefer to write, manage, and deploy your own connector code, you can use our Connector SDK instead.
The AI Connector Agent supports only SaaS sources that expose data through REST APIs and return JSON responses. It supports standard authentication methods, including:
- API key
- HTTP Basic authentication
- HTTP Bearer tokens
- OAuth 2.0
Use cases
The AI Connector Agent is ideal in the following scenarios:
- Fivetran does not already have a connector for your source, but your source exposes data through REST APIs and has publicly available API documentation that the agent can analyze.
- You do not want to wait for Fivetran to develop a connector for your source or add support through our By Request program.
- You prefer not to write connector code from scratch.
- You do not need custom connector logic.
- You use custom entities, objects, or tables specific to your implementation and want to sync that data into your destination.
- You need to sync new tables, fields, or endpoints quickly and do not want to wait for native connector support.
Key benefits
Some of the key benefits of using the AI Connector Agent are:
- Accelerates time to value by generating connectors in minutes instead of requiring lengthy custom connector development
- Reduces engineering effort by generating reliable connectors without requiring teams to write, maintain, or deploy connector code
- Provides a Fivetran-managed deployment experience for generated connectors
- Supports API pagination and nested endpoints to ensure robust data extraction
- Provides precise error handling and logs for easier troubleshooting
Features
Connectors generated by the AI Connector Agent support the following core Fivetran features:
| Feature Name | Notes |
|---|---|
| Capture deletes | Supported if the source API provides deletion data. |
| Custom data | Supported if the source API provides custom objects, tables, or fields. |
| Data blocking | Supports column-level and table-level blocking. |
| Column hashing | |
| Re-sync | Supported at the connection level. |
| Row filtering |
In addition to these core features, the connectors support standard Fivetran capabilities by default, including:
- Table selection
- Incremental syncs
- Primary key de-duplication
- User-facing logs
Supported deployment model
Connectors generated by the AI Connector Agent support the SaaS Deployment model. The agent generates the connector and executes the required API calls within the Fivetran SaaS environment.
Setup guide
Follow our step-by-step AI Connector Agent setup guide to generate your connector and connect your data source with Fivetran.
Data storage
During connector generation, optimization, and sync execution, Fivetran processes and stores API responses and connector metadata in infrastructure hosted in the US East region. Before connecting production systems, verify that your use of the AI Connector Agent aligns with your organization's data residency, security, and compliance requirements.
Operational considerations and shared responsibility
When you use the AI Connector Agent, you share responsibility with Fivetran to ensure the connector functions as intended. We support the connector's operation within our platform, and you are responsible for customizing and validating the connector output as needed.
Fivetran responsibilities
Fivetran provides support for issues related to platform execution and service functionality, including:
- Connector failures: We ensure that the connector runs successfully on the Fivetran platform and provide support if it fails to do so.
- Service-related errors: We resolve issues originating from our services that impact the connector.
- Sync errors: We support issues related to sync failures, unexpected data transformations, and authentication errors within the connector’s default configuration.
- Platform-level issues: We address any platform-related problems that affect the connector’s operation.
- Initial operation: We ensure the connector works correctly during its initial operation.
Your responsibilities
You are responsible for the following aspects of the connector:
- Data validation: You are responsible for validating the data that the connector syncs. Fivetran does not guarantee the accuracy, completeness, or integrity of the synced data. If you identify a discrepancy, such as missing records or mismatched values, contact Fivetran Support.
- Undocumented errors: Fivetran focuses on handling documented errors, but may offer limited support for undocumented errors based on HTTP status codes and message parsing.
- Connector efficiency and accuracy: Fivetran does not guarantee the connector's performance, efficiency, or precision beyond its ability to run on the platform.
Availability
Connectors generated by the AI Connector Agent are account-specific, and you cannot reuse them across multiple Fivetran accounts.
Limitations
The AI Connector Agent may not be able to generate connectors for all data sources. If the agent cannot generate your connector, request a Lite connector or use our Connector SDK to build one.
During connector generation, Fivetran uses the credentials you provide to call the endpoints detected by the agent and selected during endpoint review. The agent cannot build an endpoint into your connector if:
The provided credentials do not have access to the endpoint.
The endpoint does not return any data from your source.
The agent does not support the following:
- GraphQL or SOAP APIs
- Report-based data, where the API returns pre-aggregated reports instead of raw transactional objects
- API responses in XML or CSV format
- Downloadable ZIP file responses
- Dynamic schemas that change based on the source configuration
The agent supports up to 500 endpoints per connector. If your API documentation contains more than 500 endpoints, provide an API reference PDF or OpenAPI specification that includes only the endpoints you want the connector to sync.
AI-generated connectors for the same source may have different schemas or configurations. If you need consistent schemas and configurations across multiple connectors for the same source, contact Fivetran Support.
AI-generated connectors do not support History Mode.
Additional support
If you have questions about the AI Connector Agent or need assistance generating or configuring a connector, contact our support team.