> ## Documentation Index
> Fetch the complete documentation index at: https://dadocs.metazense.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Capabilities

> The dazense chat interface provides powerful built-in capabilities for natural language analytics.

## Key Features

* **Natural Language Queries**: Ask questions in plain English and get instant answers
* **Data Visualization**: Automatically generate charts and graphs from your data
* **Transparent Reasoning**: See the SQL queries and logic behind each answer
* **Conversation History**: Navigate through past queries and build on previous answers
* **Governance Modes**: Run with no governance, semantic guidance, or full trusted execution

## How It Works

The agent processes questions through six sequential steps:

1. **Context Discovery** - Searches and reads context files to understand the data model, business definitions, and rules
2. **Execution Planning** - Chooses `query_metrics` (semantic) or SQL path
3. **Governance Check** (Trusted V1/V2) - Builds a contract and runs policy checks
4. **Execution** - Runs approved SQL/metrics with scope and policy constraints
5. **Visualization** - Creates charts when appropriate to display data
6. **Response** - Summarizes findings in plain language with sources and reasoning

## Agent Tools

The dazense agent includes tools for both analytics and governance:

* **build\_contract** - Creates and validates an execution contract before governed runs
* **query\_metrics** - Executes semantic-model metrics for deterministic answers
* **execute\_sql** - Executes SQL queries against connected databases
* **get\_business\_context** - Retrieves applicable business-rule guidance
* **display\_chart** - Creates visualizations from query outputs
* **search / list / read / grep** - Context retrieval and file navigation

The agent automatically orchestrates these tools based on your question without requiring manual tool specification.

## What changes by mode

* **Direct Analytics**: fast SQL-focused experience
* **Semantic Mode**: governed metrics and business guidance, lighter enforcement
* **Trusted V1/V2**: contract-first execution, policy checks, and provenance artifacts