Nexadata Connector Copilot

The Nexadata Connector Copilot allows you to connect to any external system with an OpenAPI specification and extract data into a Nexadata dataset, without writing code. This article walks you through the full process, from launching the copilot to finalizing your dataset columns.

With over 20,000 public APIs available on directories like APIs.guru, no two connections look exactly the same. Connector Copilot dynamically adjusts its interface based on the OpenAPI spec you provide, so the fields, authentication options, and configuration parameters you see will vary depending on the API you are connecting to. Some platforms like Google BigQuery also extend the standard OAuth flow with additional parameters specific to their platform. Connector Copilot supports these specializations natively, so you can connect to major platforms even when they go beyond the OpenAPI standard.

Step 1: Launch the Connector Copilot

From the Datasets page, click the AI Copilot button in the top right corner to open the Connector Copilot.

The Datasets list with a purple AI Copilot button beside Create Dataset, above a Deals dataset on a HubSpot Connection.

Step 2: Load Your OpenAPI Specification

Connector Copilot supports three ways to load your API specification. Select the option that matches what you have available:

Existing Data Source: connect to a data source you have already configured
Upload API Specification: upload a YAML or JSON file directly from your computer
API Specification URL: paste in a public URL pointing to your OpenAPI spec

An animation cycling the three load options, with the API Specification URL tab revealing a URL field and a Validate OpenAPI Specification button.

Once the spec is validated, click Start Building Connection.

A green OpenAPI Spec URL Verified banner under a HubSpot spec URL, with Name reading Deals and Version locked to v3.

Step 3: Choose an Authentication Method

Connector Copilot will detect the authentication options available in your OpenAPI spec. Select the method that applies to your data source:

OAuth: for providers that use OAuth 2.0 authorization flows
API Key: for providers that issue static API keys
Basic Auth: for username and password authentication
Private Apps: for platforms that use private application tokens

Build New Connection is selected, offering three detected methods: oauth2, private_apps and private_apps_legacy.

Step 4: Fill In Your Connection Credentials

Depending on your selected authentication method, complete the required fields. For OAuth connections, this includes:

Name: a label for this connection
Client ID: from your registered OAuth application
Client Secret: from your registered OAuth application
OAuth Scopes: the permissions required for the data you want to extract

Name holds HubSpot, the Client ID is visible and the secret masked, and two OAuth scopes are chosen while both token fields stay empty.

Once the fields are populated, connect to the API service based on the selected authentication method.

Both token fields are still empty, with the cursor on a green Connect via OAuth button and Start Building Dataset greyed out.

Once all required fields are complete, click Start Building Dataset at the bottom of the form.

Both token fields now hold masked values, and a green shield sits beside a Start Building Dataset button that has turned blue.

Step 5: Describe the Data You Want

On the Data Source Selection screen, describe the data you want to extract in plain language. Be as specific as you need - including field names, filters, or other context - to help Connector Copilot select the right API endpoint.

A free-text box reading Deals data with dealname, deal description and amount, above Submit and an Auto-Suggest button.

Click Submit to continue, or click Auto-Suggest to let Connector Copilot recommend a description based on your spec.

Step 6: Select the API Endpoint

Based on your description, Connector Copilot returns a list of relevant API endpoints from your spec. Review the options and select the endpoint that best matches the data you want.

Three candidate endpoints, each with its path and description: a deals search, a batch read by record id, and a paged read.

After selecting an endpoint, Connector Copilot analyzes it in combination with your prompt to determine how to structure the dataset.

A purple panel reading Analyzing Endpoint and Building plan for data extraction, with a part-filled progress bar below.

Step 7: Refine the Dataset Request

Once the analysis is complete, you can refine the request parameters based on what the endpoint supports.

Archived is unchecked, Limit holds 10, and Properties carries three chips, dealname, description and amount, above Create Dataset.

When your parameters are set, click Create Dataset.

Step 8: Advanced Configuration Access

Power users and those with API-level knowledge of their source systems can access the advanced configuration settings directly within the Connector Copilot Dataset Builder. This provides full visibility into how Copilot is extracting data from an external system, including how nested objects are flattened. Intelligent defaults handle the vast majority of cases, but edge cases that require adjustments are now fully accessible without abandoning the Copilot experience.

An expanded Advanced Config panel with JSONPath set to $.results, Flatten Max Depth 3, and the next-page token at $.paging.next.after.

Step 9: Define and Confirm Dataset Columns

Connector Copilot returns the fields available from the endpoint and maps them to dataset columns. Review each column and use the Ignore toggle to exclude any fields you do not need.

Ten returned columns with an Ignore toggle each, switched off for id, properties.amount and properties.dealname and on for the other seven.

Your dataset will be created and available in Nexadata for use in pipelines, transforms, and workflows.