All integrations

The AI data layer for Pigment

Connect Pigment to the systems your data lives in

Nexadata sits between your source systems and Pigment, so the data landing in your models arrives connected, clean and mapped. Read from Pigment, write back to it, and reach any source system on either side.

At a glance

Into Nexadata

  • Saved Pigment views
  • Blocks: Metric, Table, List
  • Payload Builder for custom JSON

Back into Pigment

  • Metric blocks
  • Dimension Lists
  • Transaction Lists

Authentication

Connect with your Pigment API key.

  • 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 Pigment connector does

Import from views, blocks or a custom payload

Pull data into Nexadata from a saved Pigment view, directly from a Metric, Table or List block, or through the Payload Builder when you need a fully custom JSON import definition. Application, block and view dropdowns filter against each other, so an invalid combination is not offered in the first place.

Write back to the block types Pigment expects

Send transformed data into Pigment as a Metric, Dimension List or Transaction List. You pick the application, block type, block and view when you configure the pipeline output, and the same dependent filtering keeps the target valid.

Column validation before anything lands

On writeback, Nexadata checks your pipeline output columns against the structure of the target Pigment block. If they do not align it stops and names the header that does not match, rather than writing a partial or malformed load into your model.

Feed it documents, not just databases

Plenty of the numbers a plan depends on arrive as a PDF invoice or a formatted workbook rather than an API response. Nexadata reads the tables out of those, including ones that run across page breaks, and lands them in Pigment as ordinary rows. You describe the layout once and every later document in that shape follows it.

Ask for the integration in plain language

Point Claude, ChatGPT or any MCP-capable agent at Nexadata's MCP server and the build happens in conversation rather than in a form. It stands up the workflow, attaches the datasets, and wires the pipelines and mappings that land data in your Pigment blocks. Running it is a request too. Throughout, the agent works on metadata alone: your data is never processed or moved by the model.

Sources

Connect what feeds it

Pigment is the destination. Everything below can feed it, and you are not limited to a fixed catalog: 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.

Workday NetSuite Sage QuickBooks Salesforce HubSpot ADP BambooHR Snowflake Databricks BigQuery Amazon S3 SFTP Excel & Google Sheets CSV PDF documents

How it works

From source system to Pigment

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

  1. 01

    Connect

    Create the Pigment connection with one API key, then point the Connector Copilot at your source systems. Anything with an OpenAPI or OData spec is reachable without a prebuilt connector.

  2. 02

    Transform

    Join, filter and reshape source data into the grain your Pigment model expects, described in plain language to the Transform Copilot rather than written as code.

  3. 03

    Map

    Align source values to the dimension members and hierarchies already in Pigment, so the load matches your model instead of forcing the model to match the load.

  4. 04

    Review

    Check the pipeline before it runs, and see who changed a mapping group, what changed, how and when. Column mismatches surface here rather than inside Pigment.

Partnership

A Pigment technology partner

Nexadata is listed in Pigment's partner directory as a technology partner, alongside the SOC 2 Type II, enterprise SSO and RBAC posture that listing records.

View the Pigment listing
“Meaningful time and cost savings, and true self-sufficiency for our business users.”
Managing Director at a Pigment Implementation Partner

Questions

Pigment integration FAQ

Does Pigment support custom integrations through an API?
Yes. Pigment exposes an API, and Nexadata connects to it with a single API key carrying the Metadata, Import and Export scopes. For imports that do not fit a saved view or a standard block, the Payload Builder accepts a fully custom JSON import definition.
Which Pigment API scopes does Nexadata need?
Metadata, Import and Export. All three are required whichever operations you plan to use, and the connection test tells you if your key is missing one before you save it.
Can Nexadata read data out of Pigment as well as write into it?
Both directions. A Pigment-sourced dataset imports from a saved view, a Metric, Table or List block, or the Payload Builder. A Pigment pipeline output writes back to a Metric, Dimension List or Transaction List block.
Do I need a prebuilt connector for my source system?
No. The Connector Copilot reads any OpenAPI or OData specification and builds a working connection from it, so you are not waiting on a connector to be added to a catalog before you can feed Pigment.
Can I build a Pigment integration with Claude or ChatGPT?
Yes, through Nexadata's MCP server. Any MCP-capable client can create the workflow, add the datasets, build the pipelines and mappings that write into your Pigment blocks, and trigger the run when you are ready. What the agent sees is metadata: connection names, block and column schemas, run state. Your data itself is never processed or moved by the model.
Can I load unstructured and semi-structured data into Pigment?
Yes. Supplier invoices, bank and brokerage statements, board reports, and workbooks carrying a logo and three title rows above the data are all legitimate inputs. For a PDF, Nexadata finds the tables inside the document, stitches back together any that run across a page break, and saves the decisions you make about them as a template, so next quarter's document in the same layout runs straight through without repeating the setup. For a workbook you point at the sheet and the cell where the real data starts, and the banner rows above it are ignored. Length is not a constraint either: a 600-page quarterly report is as valid an input as a two-page invoice.
What happens if my columns do not match the Pigment block?
Nexadata compares the pipeline output columns against the target block structure and stops the writeback if they do not align, naming the header that is missing or incorrect. Nothing partial is written into your model.

See it on your data

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