All integrations

No prebuilt connector required

Connect to any OpenAPI or OData source

Build a native connector in minutes, not sprints. Point the Connector Copilot at a spec, describe the data you want in plain language, and get back a dataset your pipelines can use.

At a glance

Read-only

Load a spec from

  • An uploaded YAML or JSON file
  • A public specification URL
  • A data source you already configured

Authenticate with

  • OAuth
  • API key
  • Basic auth
  • Private apps

What you get back

A typed, flattened dataset, ready for pipelines and transforms.

  • SOC 2 Type II
  • Zero data retention
  • SSO / SAML / RBAC
  • Cloud-agnostic

The platform processes your data. An LLM never touches it.

The connector

What the Connector Copilot does

Load the spec, however you have it

Upload a YAML or JSON file, paste a public URL, or point at a source you already configured. Connector Copilot parses the specification and validates it before you go any further, so a malformed spec fails at the first step rather than three screens later.

Authentication that adapts to the source

The form is built from the spec rather than from a template: OAuth, API key, basic auth or private apps, showing only what your source actually supports. OAuth also takes custom key-value parameters, which is what makes platforms that extend the standard, like Google BigQuery, work rather than silently expiring after an hour. Credentials stay in Nexadata's secrets manager.

Describe the data, not the endpoint

Say what you want in plain language, with as much detail as you like about fields and filters. Connector Copilot searches the spec, returns the endpoints that match, and analyzes the one you pick to work out how the dataset should be shaped. Claude does the reasoning; the spec and the data stay with Nexadata.

A dataset, not a payload

Nested objects are flattened and mapped to typed columns for you to review, with an ignore toggle for the fields you did not ask for and an API-level advanced view for the edge cases that need it. What you end up with is ready for pipelines, transforms and workflows.

When the shape is wrong, it fixes the shape

APIs return what suits the API, not what suits a table: objects inside objects, values spread across rows that belong in columns. The preview asks whether the data looks right, and if it does not, Detect Transformations works out what reshaping is needed and applies it. Flattening a nested response is a button rather than a project.

The catalog stops being the limit

Every integration platform has a list, and your systems are either on it or you are waiting for a roadmap. Connector Copilot replaces the list with a rule: if it publishes a spec, you can read from it. These connections are read-only, which is exactly what makes one safe to stand up in minutes against a system you do not administer. Writing back is what the native connectors are for.

Four steps from any spec to a trusted dataset. Load Spec: bring your OpenAPI or OData, upload a file or paste a URL, and Connector Copilot parses it automatically, with over 20,000 public APIs supported. Authenticate: OAuth, API key or basic auth, with Copilot detecting the methods in your spec and adapting the form to match, and credentials staying in Nexadata. Describe Data: tell Copilot what you want in plain language and it searches your spec and returns the right endpoint, powered by Claude reasoning. Build Dataset: review the mapped columns, flattened and type-mapped, ready for transforms and workflows.

Sources

What you can reach

Examples, not a catalog. None of these has a native Nexadata connector and none needs one, because each publishes a spec the Copilot can read: if a system has an OpenAPI or OData spec, the Connector Copilot reads it and builds a working connection, with no prebuilt connector to wait on.

Coupa Workday NetSuite SAP Microsoft D365 Google BigQuery Snowflake Databricks Blue Yonder Kinaxis e2open Log-hub ADP BambooHR Monday Your own internal APIs

How it works

From spec to dataset

Connect, transform, map, review. Each step is guided by a no-code copilot, and you approve the plan before anything runs.

  1. 01

    Connect

    Load the spec, pick the authentication method it declares, and let Copilot build the connection form to match the source rather than to a generic template.

  2. 02

    Transform

    Once the dataset exists it behaves like any other. Join it, filter it and reshape it in plain language with the Transform Copilot.

  3. 03

    Map

    Align the values coming back from the API to the dimensions your target model expects, including conditional logic.

  4. 04

    Review

    Inspect the proposed columns, drop the ones you do not need, and approve the plan before anything runs.

Questions

Connector Copilot FAQ

What do I need to connect to a system with no prebuilt connector?
Its OpenAPI or OData specification, as an uploaded YAML or JSON file or a public URL. Over 20,000 public APIs are documented this way on directories such as APIs.guru, and most internal services generate a spec too.
Which authentication methods are supported?
OAuth, API key, basic auth and private apps. The options shown are the ones your spec declares rather than a fixed list. OAuth also accepts custom key-value authorization parameters, which is what makes sources that extend the standard work properly instead of authenticating once and breaking when the token expires.
Do I need to know which API endpoint I want?
No. Describe the data in plain language and Connector Copilot searches the spec and returns the endpoints that match, then analyzes the one you choose to work out how to structure the dataset. There is an advanced view if you do know the API and want to see exactly what it is doing.
Can Connector Copilot write data back to the source system?
No. Connections it builds are read-only, bringing data into Nexadata as a dataset. That is deliberate rather than a gap: read-only is what makes it reasonable to stand up a connection in minutes against a system you do not administer. Writing back is what the native connectors are for, and Anaplan, Pigment, HubSpot and Salesforce all support it.
Does my data go to the AI model?
No. The reasoning runs on the specification and your description of what you want, not on your records. The platform processes your data and an LLM never touches it.
What do I get at the end?
A dataset with nested objects flattened into typed columns, which you review and trim before saving. From there it behaves like any other Nexadata dataset in pipelines, transforms and workflows.

See it on your data

Start free with your first use case, or talk to us about your stack.