Analytics Agent is a web platform for interactive visual analytics: fast dashboards and reports from your data, with deep drill-down, AI right in the interface and built-in Python. It resembles Grafana, but the focus is not on monitoring — it is on real analysis: here you don’t just watch metrics, you dig deeper — filter, cross-check, calculate, ask the AI.
Who it is for
Analytics Agent is made for analysts and teams who find “look-only” dashboards not enough. The product was created by an analyst with more than 25 years of experience in data and visualization — it shows in the details: from the block grid to server-side histogram calculations.
System requirements
Any Linux host with Docker: the app deploys as a single container, and its data lives in persistent folders — reports and settings in one, service databases in another — while Python for Jupyter Notebooks and your own scripts is already built into the image. Step-by-step instructions are in the Installation section.
First sign-in
The login and password are set by the ANALYTICS_APP_ADMIN_LOGIN and ANALYTICS_APP_ADMIN_PASSWORD environment variables (the variable holds the SHA-256 hash of the password). Without them, signing in is impossible. Sessions work through JWT tokens with a 90-day lifetime.
When the app is deployed on a server through Xedant MultiAgent with central sign-in, no separate password is needed: you sign in to the server once, and Analytics Agent opens right away. All access methods are described on the Sign-in & Authorization page.
The interface
- Documents — the report tree with breadcrumbs for nested folders, search across documents and templates as you type, recent history;
- Templates — a library of 257 ready-made analysis templates: forecasts, segments, scoring and an analytics course (details in Analysis Templates);
- Data Sources — SQLite, PostgreSQL, ClickHouse and your own Python scripts;
- Lakehouse — a file browser for your data: Parquet, SQLite, Markdown;
- Alerts — rules, contact points, histories and silences;
- Scripts — the state of the environment and the run journal of your Python scripts (details in Python Scripts);
- at the bottom of the sidebar — active agent chats and the prompts queue; settings (font, sign-out) live in the gear menu.
The interface language (English or Russian) and the font size are chosen in the settings, where the dark or light theme also switches and the license status is visible. Sound notifications about new agent messages can be turned off with one button in the header.
Your first report
- create a document with the “New document” button;
- add blocks: a chart, a table, a metric — and drag them around the grid;
- connect a data source in the “Data Sources” section;
- write a SQL query in the block configuration — or ask the AI to do all of it for you in plain language.
A report can be made interactive: put filters and what-if fields into the header, enable click-to-drill, assemble tabs for different readers and export the finished result to PDF. Details in the Reports & Documents section.
What’s next
How documents work is described in Reports & Documents, what they are made of — in Block Types and Charts. Connecting databases — in Data Sources, how the AI works — in the agent in the report, ready-made solutions — in Analysis Templates.