Agentic User

The Agentic User is any business professional who interacts with data through natural language conversation rather than through SQL or specialized tools. The MATIH Platform makes the Agentic User the primary persona -- the conversational AI interface is the default interaction model for the entire platform.

Role Summary

Attribute	Details
Primary workbench	Agentic Workbench (3003)
Key services	AI Service (multi-agent orchestrator), Query Engine, BI Service
Common tasks	Ask data questions, explore results, create visualizations, share findings
Technical depth	Low -- natural language only, no SQL or coding required

Day-in-the-Life Workflow

Time	Activity	Example Interaction
9:00 AM	Check morning KPIs	"Show me today's revenue compared to yesterday"
9:30 AM	Investigate an anomaly	"Why did customer churn increase last week?"
10:00 AM	Prepare for a meeting	"Create a chart of Q4 sales by region"
11:00 AM	Deep dive analysis	"What factors correlate with high customer lifetime value?"
1:00 PM	Share findings	"Save this chart to the executive dashboard"
2:00 PM	Follow-up question	"Now break that down by product category"
3:00 PM	Export results	"Export this analysis as a PDF report"

Key Capabilities

Conversational Analytics

The AI Service orchestrates a multi-agent pipeline that transforms natural language into insights:

Feature	Description
Natural language queries	Ask questions in plain English
Context-aware conversation	Follow-up questions that reference previous context
Multi-intent support	Combines data retrieval, analysis, and visualization
Session memory	Maintains conversation history within a session
Streaming responses	Real-time token streaming via Server-Sent Events

AI Agent Pipeline

Each query passes through a pipeline of specialized agents:

Agent	Role
RouterAgent	Classifies intent (SQL query, analysis, documentation)
SQLAgent	Generates SQL using schema metadata and RAG context
AnalysisAgent	Interprets query results and generates natural language insights
VizAgent	Recommends chart type and configuration
DocumentationAgent	Answers questions about data definitions and business terms

Visualization

The platform automatically generates appropriate visualizations:

Chart Type	When Used
Line chart	Time-series data with trends
Bar chart	Categorical comparisons
Pie chart	Proportional distributions
Table	Detailed row-level data
KPI card	Single metric with comparison
Map	Geographic data

Conversational Flow Example

User:  "What was revenue last quarter?"
Agent: RouterAgent classifies as SQL_QUERY
       SQLAgent generates: SELECT SUM(amount) FROM orders
                           WHERE order_date >= '2025-10-01'
       Query Engine executes via Trino
       AnalysisAgent: "Revenue last quarter was $2.4M,
                       up 12% from the previous quarter."
       VizAgent: Recommends bar chart comparing quarters

User:  "Break that down by region"
Agent: Uses session context to modify query
       SQLAgent: SELECT region, SUM(amount) FROM orders
                 WHERE order_date >= '2025-10-01'
                 GROUP BY region
       VizAgent: Recommends horizontal bar chart by region

Backend Services

Service	Port	Interaction
`ai-service`	8000	Multi-agent orchestrator, conversation management
`query-engine`	8080	SQL execution for AI-generated queries
`catalog-service`	8086	Schema metadata for query generation
`semantic-layer`	8086	Business metric definitions
`bi-service`	8084	Dashboard integration for saving charts

Related Chapters

Conversational Analytics -- Full capability description
Agent Workflows -- AI agent architecture
Agent Flow -- End-to-end agent execution flow

ML Engineer Platform Admin