Skip to content

What is Polaris

Polaris Data Analytics, or Polaris for short, is an AI data analyst. You ask a question in plain language. Polaris finds the right tables, writes SQL, runs it against your database, and answers with text, tables and charts. You can collect results into stories, share them with your team, and keep them up to date on a schedule.

  1. It reads the project context. Each Polaris project has a folder of context: rules written by your team, documentation for every table, optional business docs, and a semantic layer. The agent reads this context before it writes SQL.
  2. It queries the data. The agent runs read-only SQL against the project’s databases and pages through the results.
  3. It shows the result. The agent explains what it found, and it can draw a chart, build a table or write a story.
  4. It suggests next steps. After most answers, Polaris offers one to three follow-up questions.

When a question is ambiguous, Polaris asks you to clarify before it runs anything.

Task Where to read more
Ask questions and refine them with follow-ups Chats
Look at SQL, export results, and edit charts Charts and tables
Build reports that refresh themselves Stories
Let Polaris remember your preferences Memory
Look things up on the public web Web search and fetch
Attach files such as CSV, Excel or PDF File attachments
Share chats and stories, and rate answers Sharing and feedback
Ask Polaris in Slack Slack
Use Polaris from Claude Code, Cursor or Codex MCP access
  • Users chat, build stories and share results.
  • Context owners maintain the project folder: RULES.md, table docs, skills and tests. See Project context.
  • Admins add users, choose models, and watch usage and costs. See Administration.

Polaris runs on Amazon Bedrock AgentCore and uses Claude models on Amazon Bedrock. Read the platform overview for a short description.