Skip to main content
Validio AI features use large language models (LLMs) and machine learning to help you set up monitoring, investigate incidents, and describe and connect your data. You can use AI in Validio in two ways: through features built into Validio, which run on an LLM credential you configure, and through the Validio MCP Server, which lets your own AI assistant work with Validio.

​
Enabling AI features

AI features built into Validio call your LLM provider with a credential that you configure. To set one up:
  1. Turn on Allow LLM Credentials under Workspace > Settings > AI features. This requires global settings:WRITE permission. See Managing Workspace Settings.
  2. Add a credential for your LLM provider. See LLM Credentials.
Some features also work without a credential, using heuristics or statistical methods instead:

​
AI features in Validio

AI Description Generation

Generate contextual descriptions for catalog assets, schema fields, glossary terms, domains, validators, and lineage edges using your LLM credentials.

Debug Query Suggestions

Ask an agent to write an investigation query for an incident, drawing on the incident’s details, catalog metadata, and your saved debug queries.

Generating SQL in Validators and Filters

Compose and correct the SQL queries in custom SQL validators and SQL filters with an LLM, from a description in plain language.

Glossary Term Suggestions

Use AI to automatically discover catalog assets and schema fields that match a glossary term, using text matching, lineage analysis, and LLM refinement.

Lineage Edge Suggestions

Automatically suggest field-level lineage connections between cross-system assets using heuristic and LLM-powered matching.

Validator Recommendations

Streamline validator setup by automatically analyzing your data source to suggest tailored validators, windows, filters, and segmentations—powered by LLMs or statistical ML.

​
Validio MCP Server

The Validio MCP Server lets an AI assistant of your choice, such as Claude Desktop or Cursor, work with Validio through the Model Context Protocol. Your assistant brings its own model, so the MCP Server needs no LLM credential. Instead, an administrator turns on Enable MCP Server under Workspace > Settings > AI features.

MCP Server Setup

Connect MCP clients directly to your Validio platform to help you configure validators, analyze incidents, and manage data quality, all through natural language conversation.
For details on what data AI features send to a Large Language Model, and how the MCP Server’s trust model differs, see AI Data Transparency.