The context layer that steers agents.

Metrics, instructions, and domain experts live next to the data, so every agent answer stays on-definition.

Start free trial Book a demo

Business KPIs and aggregations defined in natural language.

Sum of sales_amount for completed transactions in the selected period.

1
SELECT SUM(sales_amount) AS total_revenue
2
FROM retail.fact_sales
3
WHERE fiscal_year = 2025;

Domain context

  • Certified metrics, agent instructions, and domain experts live together, one source of truth per domain.

Glossary terms

  • Define business language once, then attach terms to columns so agents interpret fields consistently.

Query history

  • Parse past SQL into reusable templates and let the agent retrieve proven queries during chat.

Dataset metadata

  • Descriptions, owners, and tags sit on the table, not in a slide deck the model never sees.

Rules and skills at the user level

Write instructions the AI follows when querying on your behalf. Public rules can be cloned by teammates in your organization.

  • Tenant filter

    Public

    Always filter by company_id

  • percentage formatting

    Public

    Always format percentages between 0 and 100 with % as suffix.

Add glossary terms and organize them in groups

Group related definitions under a shared domain, browse contents, documentation, and owners in one place.

Chicago Crime

Term Group · 2 items

ContentsDocumentationOwners

Case Number

Identifier assigned to a police case. It is stored as text and may not always be unique, so use report_no when uniquely identifying an individual incident.

Crime Incident

A recorded event involving a reported criminal offense. Each incident is uniquely identified by report_no.

Attach terms to columns

Link glossary terms to schema fields so agents and analysts read the same business meaning from every column name.

region

Geographic sales region (North America, EMEA, APAC, LATAM).

VARCHAR
Sales Region

sales_amount

Total Revenue
Attach term
Sales Region
Sales Region Sales
Total Revenue Sales
Fiscal Quarter Finance

Parse SQL into templates

Turn a validated query into a parameterized template with summary, purpose, and analytical follow-ups.

What was total revenue by region for fiscal year 2025?

1
SELECT region, SUM(sales_amount) AS total_revenue
2
FROM retail.fact_sales
3
WHERE fiscal_year = 2025
4
GROUP BY region
5
ORDER BY total_revenue DESC;
1
SELECT region, SUM(sales_amount) AS total_revenue
2
FROM retail.fact_sales
3
WHERE fiscal_year = {{fiscal_year}}
4
GROUP BY region
5
ORDER BY total_revenue DESC;
Templated

Summary

Regional revenue rollup for a selected fiscal year using certified sales_amount.

Purpose

Answer executive questions about revenue distribution across sales regions.

Sales Golden Usage: 47

Reuse proven queries in chat

When a user asks something similar, Unity Agents retrieves matching history, with usage and trust signals.

User question

What was total revenue by region last year?

Matched query history

What was total revenue by region for fiscal year 2025?

Regional revenue rollup for a selected fiscal year using certified sales_amount.

Templated Golden Usage: 47
1
SELECT region, SUM(sales_amount) AS total_revenue
2
FROM retail.fact_sales
3
WHERE fiscal_year = 2025
4
GROUP BY region
5
ORDER BY total_revenue DESC;

Related

Ready to make self serve analytics a reality for your company?

Deploy Unity Layer AI agents and empower your business and data teams with fast and accurate AI insights

Get Started