Data Sources

Data sources are the foundation of your Corvic AI workflows. Corvic provides several ways to bring in your data — file uploads, live data connectors, and web search from the source menu, plus programmatic fetching from any external API — so you can work with both structured and unstructured data from virtually any source.

​

Create Data Sources

To add a data source, click the ”+” button next to “Sources” in your data room. You’ll see three options:

​

Data Connectors

Connect directly to external storage and data platforms without moving or duplicating your data. Corvic supports the following connector providers:

• Amazon S3 — Pull files from S3 buckets
• Azure Blob Storage — Connect to Azure containers
• Snowflake — Query data warehouses directly
• Google Cloud Storage — Access GCS buckets

To set up a connector, an admin must first configure the integration through the Data Connectors section in the Admin Console. Once validated, select “Data Connector” from the source creation menu, choose your provider, and specify a path:

• End the path with / to pull data from a folder
• Enter a filename (without extension) to upload a specific file
• Enter * to connect everything in a folder and its subfolders

You can set a Sync Frequency when creating a data source to automatically pull new data from your connector at a chosen interval.

​

File Upload

Upload files directly from your local system. Select “File Upload” from the source creation menu, give your data source a descriptive name, and choose the data type (Structured or Unstructured). For the full list of supported formats, see Data Types. Drag and drop your files or click “Choose File” to browse. The maximum file size per upload is 1 GiB, with a total room storage limit of 5 GiB.

​

Web Source

Search the web and extract structured data directly into Corvic — no manual data collection needed. Select “Web Search” from the source creation menu and describe what you’re looking for in plain language. Corvic’s agent will search the web using your natural language query, extract the relevant information, and create structured tables with a desired output schema. Use the Suggest Schema button to let Corvic recommend a table structure based on your query, or define your own.

​

Fetch from an API

Beyond the source-creation menu, you can pull data directly from any external API using the Python action on the Workflow canvas. The agent’s execute_python step can call a service and load the response straight into a Corvic Table — useful for live or on-demand data that isn’t a file or a configured connector. Credentials are managed through the API Secret Vault: connect a service once, then reference it by name in code. Corvic’s proxy injects the secret into the request, so your API keys never appear in the code, logs, or chat history. The catalog includes one-click presets for 35+ services (Crunchbase, PubMed, Polygon.io, and more), or you can add any custom HTTPS origin.

​

Incremental Uploads

Additional files can be added to an existing data source via the “Add Files” button. New files must match the original data source’s file type. This allows you to continuously grow your data sources as new data becomes available.

​

Related Documentation