1.2kGet started
Datus Web Chatbot

Datus Chatbot — Analyst-Grade Chat UI for Everyone

Bring the same agent that powers the CLI into Slack and the browser, so analysts and PMs get governed answers without touching a terminal.

Why a Chatbot for Analytics?

Not every stakeholder wants a SQL editor or a BI dashboard. They want to ask a question where they already work — and get an answer they can trust.

Meet Users Where They Work

Not every stakeholder wants a SQL editor or a BI dashboard. They want to ask a question in the tool they already have open — Slack, or a browser tab — and get an answer they can trust.

Trusted, Traceable Answers

Every reply cites the tables and metrics it used, with lineage back to the semantic layer. No hallucinated column names, no guessing about definitions — the same context the CLI uses.

Zero Install for Stakeholders

Add --web once. Teammates open a URL and start asking — no pip, no Docker, no login they'll forget by Tuesday.

Chatbot Use Cases Across the Company

Four conversations that used to be tickets — now they happen in the thread where the question was asked.

Answer PM Questions in Slack Without a Ticket

A PM asks “how did last week's launch perform” in #growth. The bot replies with governed SQL, a mini chart and lineage — no ticket, no context switch for the data team. The bot wraps a subagent packaged in Studio and hands business users the self-serve surface analysts already trust.

On-Call Data Diagnostics From a Slack DM

Revenue looks wrong at 8am. Ask the bot in Slack; it pulls the freshness check, spots the late job, and tags the owner — before the exec ping hits your DMs. When the trail runs deeper, root-cause continues in the CLI using the same anomaly loop analysts run.

Self-Serve Ops Runbooks in Slack

Wrap a subagent as a Slack bot pointed at ops's Skills library. Ops asks “which shipments are delayed” in plain English; the runbook answers with the same governed SQL every time — trigger the same subagent via API when the caller is a cron job instead of a person.

Close the Loop From Chat Back to Context

Analysts thumb-up the answers that worked and flag the ones that didn't. Every upvote and issue report — session link attached — flows back to the data engineer who tunes the subagent in Studio's review queue, so tomorrow's questions get the sharper answer.

chatbot.ymlyaml
chatbot:
  channels:
    - kind: slack
      workspace: T0123ABCD
      allowed_channels: [growth, exec]
    - kind: web
      port: 8501
      auth: sso
  agent:
    context: growth
    skills: [mrr, funnel, cohort]
    policy: read_only

Deploy in Slack, Web, or Both

One config, two channels. Point the Slack app at your workspace, start the Streamlit web UI, or run both against the same agent and context — every message goes through the same policy layer.

Chatbot Loop: Chat, Recall, Share

Three things the Datus chatbot does that a generic LLM chat UI cannot.

Context-Aware Chat

Answers cite the exact tables, metrics and semantic definitions used — no fabricated column names, no vague guesses.

Session Recall

Every conversation is a shareable URL with full transcript, SQL and tool calls. Reopen tomorrow and pick up exactly where you left off.

Governed Sharing

Reply gets a share button that respects workspace access policies. Teammates see the exact query and result, no screenshots.

Launch the Chatbot in Under a Minute

Same install as the CLI. Point it at Slack or the browser and hand teammates a URL.

  1. 01

    Install the agent

    Grab Datus from PyPI or Homebrew — the same binary powers the CLI, chatbot and API.

  2. 02

    Enable a channel

    Turn on the Streamlit web UI or register the Slack app — one flag, no extra dependencies to install.

  3. 03

    Share the URL

    Send teammates the workspace URL or invite the bot to a Slack channel. They start asking questions in plain English.

Pick the Interface That Fits Your Team

Four surfaces. One agent. Pick the one that fits your team.

CLI

Explore data, build context, and ship SQL from the terminal.

You are here

Web Chatbot

Chat with subagents from a browser — zero install.

API Server

Consume data services via REST — language agnostic.

MCP Server

Plug into Claude Desktop, Cursor, and any MCP client.

FAQ

Frequently asked questions

Who is the Datus Web Chatbot for?

Analysts, PMs and business users who want to query the warehouse in plain language from a browser — without learning SQL or installing a CLI.

How is it different from a generic AI chatbot like ChatGPT?

The Datus chatbot is grounded in your warehouse, catalog, metrics and team-approved answers via the evolvable context engine, so it produces governed SQL and traceable results — not hallucinations.

Is it free and open source?

Yes. The web chatbot ships with the same Apache 2.0 codebase as the CLI — launch it locally with `datus-cli --web`, or use the hosted Datus Studio.

Can I share a chat thread with my team?

Yes. Threads have stable URLs that carry the subagent, datasource and full context, so anyone on the team can reopen and continue the conversation.

Give Your Team a Browser Data Chatbot

Share one URL with analysts and PMs, let them chat with the warehouse, save subagents — no install, no SQL ramp-up required.