Skip to content

Data Connectors

Data connectors are simple yet powerful integration tool that allows you to easily integrate your AI/ML systems to your data sources. Data conncetors work by automatically retriving data should a your core AI/ML requires it for processing or making predictions. They are felxibly, powerful, highly reliable and do not require any API logic or API access. Data connectors are a from of integarion, you can use data conncetors with INTELLITHING router engine or in a manually created workflow.

🔑 Key Concepts

Data Connectors are lightweight but powerful integration blocks that allow your AI/ML systems to retrieve relevant data from external sources—just-in-time, when it's needed for inference, enrichment, or decision-making. You can add as many data connectors as needed.

These connectors act as the entry point for structured or semi-structured data across various platforms (e.g., SQL databases, Notion, Slack, Snowflake, GitHub, HubSpot, PDFs), enabling your workflows to dynamically access the latest information without manual intervention.

How Data Connectors Work

  • Triggered Retrieval: Data is retrieved on-demand, either by the INTELLITHING Router Engine or within a custom workflow. Retrieval is context-sensitive, based on user queries or internal logic.

  • Structured Input: Each connector exposes a simple configuration interface (API token, data source metadata, filters, etc.), making it easy to authenticate and target the correct source.

  • Modular Workflow Nodes: All connectors follow a standard processing pipeline:

  • Ingest Configuration – Initializes and validates connection settings.
  • Retrieve Nodes – Pulls data from the connected source.
  • Rerank Nodes – Sorts results by relevance (optional but common).

  • Synthesize Nodes – Combines or formats results for use by downstream tasks or LLMs.

  • Flexible Integration: Data Connectors can be used in:

  • Auto-generated workflows powered by the router engine.

  • Custom-designed workflows where the logic and sequence are defined manually.

  • Security and Scope: Connectors use secure tokens (e.g., Slack, HubSpot, GitHub) and often allow scoping (e.g., limiting by table, page ID, channel, file extension) to enhance performance and privacy.

Why Use Data Connectors

  • Eliminate the need for preloading data into memory or hardcoding access logic.
  • Enable just-in-time data access for real-time decision-making and analysis.
  • Avoid hallucinations and improve model accuracy by grounding responses in live, verifiable data.
  • Support scalable and modular architecture for enterprise-grade ML systems.

⚙️ Configuration

  1. Head to block editor and drop any module from the data conncetor section.
  2. Click on the block and fill in the required configuration variables.

    ℹ️ Note that every data connector block require its own uniqe configuration. You can find detailed guide by accessing their individual documentation under Data connectors.

  3. save the block
  4. Head to the workflow editor and see the default workflow.
  5. you can save the project, add LLMs and deploy or you can create your own workflow using bridges. For more information regarding worlfow see workflow documentation