Multi-agent

Give every schema a specialist

Each DataAgent specialist has a defined table scope, its own business context, and a focused set of questions it is built to answer.

E-commerce — orders + productsactive
Restaurant — reservations + menuactive
Healthcare — patients + visitsactive
Wealth — portfolios + tradesactive
Real Estate — listings + agentsdraft

Purpose-built context

The right knowledge follows the right data

An agent is more than a database connection. It packages the schema and business context needed for consistent analysis.

Table scope

Choose the database, schema, and permitted tables each specialist can query.

Knowledge base

Attach definitions, business rules, documents, and examples that explain what the columns mean.

Semantic model

Define metrics and logical relationships so recurring questions follow the same business logic.

Preset questions

Give teammates reliable starting points for the decisions they make most often.

Focused retrieval

Recall stays within the world of each specialist

Each agent retrieves its own embeddings and business context before planning SQL, keeping similar terms from unrelated teams out of the answer.

  • Per-agent embedding recall
  • Tenant and agent boundaries enforced in SQL
  • Foreign agent rows are inaccessible by direct ID
Isolation boundaryenforced
Tenant
Agent + knowledge + embeddings
Permitted tables only

Build the specialist your team is missing

Start with one schema, teach it your business language, and expand agent by agent.