DataClawChat with your data

Ask your data. Trust the answer.

Dashboards answer the questions you already knew to ask. DataClaw answers the next one, in plain English, across your Power BI models, and checks the answer against the data with obviEnce's own validation technology.

The demo runs on sample sales data. Sign in with a code sent to your email.

DataClawLive answer · sample data
A complete DataClaw answer on sample sales data: the question 'Which product categories drove our sales growth in 2025, and which held us back?', the opening story, a ranking chart, a grouped bar chart, suggested next questions and the green Data-grounded badge.

01The problem

Dashboards answer yesterday's questions.

Built in advance

Built for the questions you already knew

A report shows what someone decided to measure when it was built. The question you have today usually isn't on it.

The wait

The next question means a new request

A new page, a new export, or a wait for the analyst, while the moment passes.

The silos

The answer crosses departments

Sales, finance and operations each have their own model and their own reports. The question that matters often needs more than one.

02How it works

From a question to an answer you can check.

  1. 1Ask in plain EnglishType a question the way you would ask a colleague. Attach a file or paste an image if it helps.
  2. 2Query as youDataClaw queries your governed Power BI model with your own sign-in, so you see only what you're allowed to see.
  3. 3Tell the storyA short narrative of what the numbers say and why it matters, with the charts that make the point.
  4. 4Check the answerAn independent check compares the finished answer with the data that was actually fetched.
  5. 5Show the resultA badge shows the result of that check, and DataClaw suggests what to ask next.

Starts answering in secondsPower BIMicrosoft FabricMicrosoft sign-in

03Across your business

One question. Every area you're allowed to see.

Your sales, finance and operations numbers live in separate Power BI models. DataClaw knows each one through its knowhow, picks the models a question needs, and brings the results together in one answer.

Why did margin fall in the West last quarter?

Sales modelruns as you Finance modelruns as you Returns modelruns as you

Margin in the West fell 2.1 points. Price held; higher returns in two product lines and a freight cost rise in finance explain most of the drop.

Data-groundedIllustrative
  • 01
    SecurelyEvery query runs as you, model by model. If you can't open it in Power BI, DataClaw can't show it to you.
  • 02
    ConsistentlyEach area's definitions come from its knowhow, so a term means the same thing whoever asks. Ask for a redo and DataClaw re-runs the same queries.
  • 03
    ReliablyThe independent check applies to every chart section, whichever model it came from.

04Proprietary validation technology

Checked against the data, and honest when it can't be.

DataClaw is built to give secure, consistent, accurate answers, and to tell you when it can't verify one. The check is obviEnce's own proprietary technology: an independent review of the answer against the rows it was built from. A general AI assistant answers confidently, but often can't show where a number came from. DataClaw can.

The answer
  • West revenue+$1.4M
  • North revenue+$0.7M
  • South revenue−$0.9M
  • East revenue−$1.3M
The data
  • West1,402,118
  • North 698,540
  • South −412,906
  • East−1,291,377

South doesn't match the data. DataClaw revises that section once and checks it again. If it still doesn't match, the section is removed or flagged, and the badge says so.

1 issue found · see report Illustrative
  • 01
    A badge on answers from your dataWhen DataClaw answers from your data, a badge shows the result of an independent check against the rows it pulled.
  • 02
    Numbers traced to your dataEach chart value is traced to a row that was actually fetched. DataClaw works from the measures your Power BI model defines, and the independent check catches arithmetic that doesn't match the data.
  • 03
    Honest when it can't verifyWhen a chart or statement can't be verified, DataClaw removes or flags it and tells you. When it can't answer from your data, it says so.
DataClaw's fact-check badge opened to show its report: the data queries that ran, each chart section's values traced to a query, and the line 'Independently verified, 4 chapters and prose checked against the raw data'.
In the product. Open the badge and see the report behind it: which queries ran, and which values each chart section traces back to.

We work hard to earn your trust.

05Insight, not a dashboard

The chart says what happened. The words say why.

DataClaw surfaces the one to three things that matter, explains what they imply, and suggests what to look at next. Restyle a chart by asking, copy it as an image, or export its data.

0chart types
and growing
  • Trend
  • Waterfall
  • Ranking
  • Quadrant
  • Heatmap
  • Treemap
  • Funnel
  • Pareto
  • Scorecard table
  • Comparison
  • Breakdown
  • Scatter
  • Bullet
  • Dumbbell
  • Slope
  • Small multiples
A DataClaw waterfall chart: 2025 sales bridge from $339.5M to $425.2M through price, volume, mix, new products and discontinued products, with a one-line reading underneath.
A sales bridge, with the one-line reading that says what it means.
A DataClaw chart card with quarterly revenue and margin percentage, 2025 against 2024, side by side.
Trends side by side: revenue and margin %, quarter by quarter.

06Your expertise, built in

Answers as good as the people who know the business.

DataClaw reads your company's own definitions: which measures are the governed ones, what your terms mean, and the traps to avoid in your data. We call each one a knowhow.

01

Meaning, not data

A knowhow describes a business area: its measures, its terms and its known traps. It holds meaning, not your data.

02

A head start

A first draft can be drawn from your live Power BI model, then refined by the people who know the business.

03

Yours to edit

Your company owns its knowhows and edits them in an admin editor, with full revision history.

07Secure by design

Your permissions stay in force.

DataClaw doesn't add a second set of data permissions to maintain. It works inside the security you already have in Power BI.

  • Microsoft sign-inUsers sign in with their Microsoft Entra ID account.
  • Every query runs as the userPower BI workspace permissions and row-level security apply, so each person sees only what Power BI already lets them see.
  • Live, from your modelDataClaw queries your semantic model live for each question. There is no nightly copy or separate warehouse. Each conversation keeps its answers and the rows behind them, so you can come back to it.
  • Run by obviEnce on Microsoft AzureA hosted service, with one configured tenant per customer and an audit log.

08What you get

On top of the Power BI you already have.

DataClaw works with the semantic models you already have in Microsoft Fabric and Power BI. No separate warehouse, and no second set of data permissions to maintain.

  • 01Export anywhere. Excel and CSV, or a whole session as PDF.
  • 02Bring a file. Attach a PDF, Word, Excel or CSV file, or paste an image into a question.
  • 03Pick up where you left off. Searchable history, with the charts, and saved prompts.
  • 04Feedback built in. Rate answers and discuss them in feedback threads.
  • 05Admin pages. Users, business areas, knowhows, and an audit log.
  • 06Your brand. Your name, logo and accent colour, per company.
  • 07Light or dark. DataClaw works in both themes.
  • 08Questions it suggests. Answers end with ideas for what to ask next.
Thousandsof questions answered
0chart types, and growing
Secondsto start answering

We make the complex obvious

Ask your first question today.

Try DataClaw on sample sales data, or let us walk you through it on yours.