Skip to content
DataSquares
Start free

DataSquares AI Analyst

AI answers you can trace to your data

Ask a question and inspect the chart, SQL and evidence behind the answer. Shared metrics and your existing data permissions guide every query.

Start freeBook a demo
Read the AI Analyst documentation
  • Answers with the chart, the SQL and the evidence one click away
  • Grounded in your certified metrics and business glossary
  • An approval card before anything is created or changed

The AI answers from your governed metrics on every plan. The glossary and certification it draws on are managed in the Catalog, which is available on the Business plan. See plans

Every answer shows its work

Open the SQL under any query-backed answer, and the Evidence panel beside it: the metric and source it used, the filters and time range, how the data was read, how much of it the chart covers, and what it assumed.

It proposes, you approve

When the AI would create or change something, an approval card shows exactly what will happen. Only that change runs, after you approve it, and DataSquares reopens the item to check that it landed.

What you can do with AI Analyst

  • Ask from anywhere

    Open AI Chat beside any page, type a question into Ctrl+K or ⌘K, or choose Ask about this card on a chart. The page you are on comes along as context.

  • Split view

    On the AI Chat page, the conversation sits beside a result canvas that shows each answer as a chart, a table, its SQL or its evidence, with Save to dashboard beside it.

  • Coverage you can see

    Every chart says how much of the data it covers, and why when it is partial. View data opens the rows behind it as a table.

  • Statistics that work without AI

    Explain this change, anomaly detection, key influencers and forecasts run as statistics on the server, under your data permissions, whether or not AI is configured.

  • Pipelines from chat

    In workspaces with Pipelines, ask why a run failed, or describe a pipeline and approve one card with every step and the schedule before it is created.
Approval first
Nothing is created or changed until you approve
Evidence
Sources, metrics, filters and assumptions behind each answer

Inside AI Analyst

Answers you can check

Each answer shows how much of the data it covers, the SQL behind it and the evidence it rests on: the sources and metrics by name, the filters and time range, whether the data was read live or from the cache, and anything it assumed.

  • Steps listed as the AI works, each ticked when done
  • Exported conversations keep their evidence, so an answer can be reproduced
  • A thumbs-up on an answer that used your data guides your own later questions, by its approach, never its values

Built on your definitions

Ask how revenue is pacing and the AI evaluates your governed metric, with its target and certification, instead of writing new SQL for a KPI that already has a definition.

  • Glossary terms and synonyms matched in questions and field names
  • Certification visible to the AI
  • Create an alert rule by asking, then approve its exact condition

Nothing changes without approval

When the AI would create or change something, the approval card shows the destination, the items affected and every change, with before and after values for anything that exists. Once it is applied, DataSquares reopens the item and shows what it checked.

  • Dashboards, reports, alert rules, email schedules and calculated fields
  • Pipelines too, when your workspace has them
  • A dashboard it builds opens from the answer, ready to refine in the editor

How it works

  1. Step 1: Ask

    In AI Chat, from Ctrl+K or on a card

  2. Step 2: Check

    Chart, table, SQL and evidence

  3. Step 3: Follow up

    Suggested follow-ups, or your own question

  4. Step 4: Save or approve

    A live card on a dashboard, or an approved change

Questions, answered

Something else? Ask our team or search the documentation.

Where do the numbers come from?
From queries the AI runs against your data, through your governed metrics when one exists. The SQL behind any query-backed answer is one click away, and Evidence lists the sources, metrics, filters and assumptions it rests on.
Can it change dashboards without asking?
No. Before it creates or changes a dashboard, report, card, filter, alert rule, email schedule, calculated field or pipeline, it shows an approval card with exactly what will happen. Only that change runs, and only after you approve it.
Does it respect data access?
Yes. The AI works as the person asking. Row-level security and field masking apply to every answer, and card insights are computed under your own permissions.
Which AI does it use, and how is it paid for?
DataSquares picks the AI behind each feature, so there is no model to choose. AI features draw on your workspace’s AI credits, and administrators can see usage by feature.
Can people outside the workspace ask questions?
Only if the publisher turns on Ask AI for a share link or an embed. It is off by default, answers only from that share’s own cards, and is rate-limited per share.

Works well with AI Analyst

All products
AnalyzeDataSquares BIInteractive dashboards. Shared definitions.AnalyzeDataSquares StoriesNewTurn live data into a shared business storyGovernDataSquares CatalogA shared catalog for trusted data

See AI Analyst on your own data.

Start a free trial in minutes, or book a walkthrough with our team.

Start freeBook a demo