Show HN: Beyond "chat with your database" – agentic data exploration
reportr.com
Today, to answer a business question against a database, analysts use a tool like Tableau: they select relevant data, build an initial view using fields, filters, calculations, and visualizations, then look for patterns or anomalies. They form hypotheses and test them — filtering by region, drilling into customer cohorts, comparing time periods, or creating calculations to validate or rule out an explanation. They repeat this process until the evidence supports an answer, then save a workbook or dashboard for others to review and monitor.
With a single question, AI can now perform much of this exploratory loop: it can investigate the data, follow the evidence through multiple queries, and present the supporting analysis. This is agentic data analysis — AI autonomously decides what to investigate next based on the results it has already found.
Reportr puts agentic AI at the center of the experience, then builds complementary capabilities around it to address AI's limitations:
SEMANTIC LAYER
Production databases are rarely clean: columns that are no longer used, cryptic names like status2, misc, or flag, columns whose meaning has quietly shifted over time, tables that linger long after their purpose is gone, and so on. This is known as schema rot. AI tools guess wrong against schemas like this. Reportr's semantic layer lets you give tables and columns business-friendly names and descriptions, document relationships, capture business rules, and flag what AI should ignore — so it's working from what your data means, not just what it's called.
USER OVERSIGHT
AI-generated answers can be wrong. Reportr lets you inspect the SQL executed to arrive at an answer. It also provides explanations of what the SQL does (understandable by business users), and how it works (intended for technical users). You can review the analysis rather than simply accepting AI's conclusions.
BETTER USER EXPERIENCE
Reportr UI is specifically optimized for chatting with a database. Database queries often return large number of records. Ask which customers haven't placed an order in the last 90 days, and the answer is a list of customers. Reportr displays tabular data in an interactive grid that can handle thousand of records without breaking a sweat. The grid supports sorting, searching, and filtering.
REPORTS AND DASHBOARDS
Chat is ephemeral. Every conversation starts fresh, and that's fine for exploration — but not for information you need every week. When a chart or table reveals something worth tracking, save it as a report. Combine reports into dashboards to monitor changes over time, share findings, and keep important insights accessible without having to ask the same questions again.
ROLE BASED ACCESS CONTROL
Not everyone on your team needs the same level of access. Reportr's role-based access control lets you give each user exactly the permissions they need — from view-only access to report building, database management, and full administrative control.
CENTRALIZED MANAGEMENT
Reportr is a web application — there's nothing to install, nothing to update, and no desktop app to manage across your team. Database connection strings, AI keys, are all managed centrally. Reportr can connect to all supported databases directly, MCP servers and drivers are not needed.
INSTALL ON-PREMISE
Reportr can be self-hosted on your own infrastructure, ensuring your data never touches our servers. You can supply your own AI API key. Your data will not be used for training AI models. If you use Claude on AWS Bedrock, Amazon additionally guarantees that your data is never stored or used for training — and Bedrock is HIPAA eligible.
GIVE IT A WHIRL
Visit https://reportr.com to give it a try. We look forward to getting your feedback.
Reportr is easy to try: you can provision a sample database with the press of a button and start asking questions right away. Or connect your own database and analyze your own data. You can ask up to five questions for free.