Live charts in the run
ECharts visualizations appear inside the analysis stream, grounded in the rows just returned.
Explore live reports →Live — scheduled email reports now shipping
Ask your database anything.
Get the report, not just the answer.
DataAgent turns plain questions into governed SQL, live charts, and decision-ready reports — using your models or ours, on your schema.
Built-in AI or your own modelsSQL-enforced tenant isolationCharts and downloadable reportsMCP for your AI tools
Top product categories by revenue this quarter?
Revenue by categoryIllustration
Interface illustration — not a recorded analysis or customer result.
Intent ✓Schema recall ✓SQL ✓Validation ✓Chart ✓Report ✓
Platform
Every answer is a deliverable
A governed analysis pipeline shows its work and ships the result where your team can use it.
Scheduled email reports
Run a saved question on schedule and deliver PDF, HTML, and Markdown attachments automatically.
See scheduled reports →MCP-native access
Connect compatible AI tools with scoped, expiring access that is read-only by default.
Connect your AI →Multi-agent
One schema, one specialist
Give every database or schema its own agent, table scope, knowledge base, semantic model, and preset questions. Sales asks Sales; Finance asks Finance.
How multi-agent works →E-commerce — orders + productsactive
Restaurant — reservations + menuactive
Healthcare — patients + visitsactive
Wealth — portfolios + tradesactive
Real Estate — listings + agentsdraft
Your models
Managed or BYOK — your call
Start with dataagent-chat, or configure a compatible model under your own account and billing.
Explore model options →Managed models — dataagent-chatincluded
Your own compatible modelsBYOK
Separate chat and knowledge searchcompatible
Self-hosted models when reachablecompatible
Governed analysis
Why DataAgent
From permissioned data access to final report, every stage stays visible and under your control.
Multi-agent workspaces
One agent per data source or department — each with its own knowledge base, prompts, and permissions.
Human-in-the-loop
The AI proposes a plan; you approve before anything runs. No rogue SQL on production databases.
SQL + Python, autonomously
A 16-step agent workflow decides when to query, when to compute, and when to retry — not just chat.
Real reports, not chat logs
Get charts and HTML/Markdown reports you can hand to your team, generated from live query results.
Use your own models
Analysis with your own compatible model does not count against your managed query allowance.
MCP native
Connect compatible MCP clients directly to your agents.
Workflow
How it works
Connect your database
MySQL or PostgreSQL — read-only credentials recommended. Your data stays in your database.
Ask in plain language
"Which region grew fastest last quarter?" The agent plans, queries, and analyzes.
Get an approved answer
Review the plan, approve execution, and receive charts plus a shareable report.
Proof
From question to inbox
A saved business question becomes a governed analysis and arrives as a polished report.
Email subject
[DataAgent] Scheduled report: Monthly patient visit trends
Attachments
report.pdf · report.html · report.md
Cadence
Daily at 09:00 America/Toronto
Start with a question. End with a report.
Connect your database, create a specialist, and turn a real business question into a usable deliverable.