[Use case] Data · leadership, ops and finance teams

Ask your data a question. Get the number, and how.

Every question about the business goes to the one person who knows the database, and waits in their queue for days. For a team with 10 or more people asking for numbers each week, the answers arrive after the meeting that needed them.

Skip this if your dashboards already answer the questions people ask. Better dashboards will do.

What you get
Plain-English questions answered from your database and CRM, with the query shown
Pilot length
4 to 6 weeks on one data source
Price range
Pilot from US$2,400, version 1 from US$7,200
We need from you
Read access to your database, your key definitions, and one analyst to check answers
[01] What it doesFour jobs, one question box

Asked, answered, shown.

01Answers in plain English

Ask the way you'd ask a colleague, and get the number, a chart and one line on what it means.

02Shows its working

The query it ran and the tables it used, so an analyst can check any answer in seconds.

03Uses your definitions

Revenue, active customer and churn mean what your team says they mean, not what the model guesses.

04Writes the weekly summary

A short report on the numbers that moved, sent to Slack or email every Monday.

Web app · ask your data
app.yourcompany.com/ask-data
AnswerJul to Sep 2026 · 2.1 s
38customers churned

Most of the rise is on Starter, after the August price change. Pro and Team stayed flat.

Starter
24
Pro
9
Team
5
Query it ranRead-only
SELECT plan, COUNT(*) AS churned
FROM customers
WHERE churned_at >= '2026-07-01'
  AND churned_at < '2026-10-01'
GROUP BY plan;
Your definitionChurnedNo paid invoice for 60 days. Agreed by finance, 3 Mar.
Board packNumbers for the board go to an analyst to sign off first✓ Signed off
Concept · sample data
[02] How it worksInputs, model, actions, review

One flow, with an analyst checking.

The AI writes the query and the summary. Your definitions decide what the words mean, and an analyst signs off anything that goes in a board pack.

01 · Inputs

What comes in

Questions from Slack or the web app, plus your database, CRM and the definitions your team agreed.

QuestionDatabaseDefinitions
02 · Model

What it works out

Turns the question into a query, runs it read-only and explains the result in a line.

QueryRunExplain
03 · Actions

Where it puts it

The answer, chart and query go back to whoever asked, and the weekly summary goes out on Monday.

SlackWeb appEmail
04 · Review

Who checks it

An analyst checks a sample of answers each week and approves new definitions.

CheckApprove
Ask your data · demo Sample data

The demo needs a bigger screen, so it opens full screen on a phone.

QuestionKim · finance

How many customers churned last quarter, by plan?

  1. Definition

    Churned: no paid invoice for 60 days. Agreed by finance, 3 Mar.

  2. Query

    SELECT plan, COUNT(*) FROM customers WHERE churned in Q3 GROUP BY plan

  3. Answer

    38 customers churned: Starter 24, Pro 9, Team 5. Most of the rise is on Starter, after the August price change.

Answered

The answer, a chart and the query go back to Kim in Slack.

Confidence 0.93 · bar 0.85
[03] Build pathTimes and prices from past builds

Pilot, then version 1, then run it.

Each step is priced on its own, so you can stop after the pilot if the answers aren't right often enough.

01

Pilot

One data source and the 20 questions your team asks most, with an analyst checking every answer.

Time
4 to 6 weeks
Price
From US$2,400
Ends with
A go or stop, with the share of answers the analyst passed
02

Version 1

Your main sources, Slack and web, the weekly summary, and a log of every query.

Time
5 to 6 weeks
Price
From US$7,200
Ends with
Answers people trust without asking an analyst
03

Run and improve

Model updates, new sources and definitions, and monthly tuning on the answers that were wrong.

Time
Monthly
Price
From US$599 a month
Ends with
A report every month

In US dollars · monthly plans on the pricing page

[04] EstimateTwo minutes, before any call

Estimate this build.

Answer three questions and the estimator opens with ask your data already picked.

Typical · pilot from US$2,400 · version 1 from US$7,200
People asking
Where the data lives
Sources to connect
Open the estimator
[05] Buy or buildThe tools that already do some of this

You may not need us. Here's how to tell.

BI tools and chat assistants now answer data questions. If one covers what you need, use it. These are the usual options and where each one stops.

ToolWhat it coversWhere it stops

Power BI Copilot

Answers questions and builds visuals from data already modelled in Power BI.

Only as good as the Power BI model behind it, inside Microsoft's stack.

ThoughtSpot

Search and AI answers over your warehouse, on a semantic layer you set up.

A full analytics platform your team has to set up and run.

ChatGPT data analysis

Upload a spreadsheet and ask questions about it.

Works on files you upload or attach; live databases need connectors an admin sets up.

When a custom build winsAnswers from your live database, definitions your team agreed, or numbers that must show the query behind them.

That is where an off-the-shelf BI assistant runs out, and where this build earns its price.

[06] Data, security and guardrailsWhat it can and can't touch

It reads. It never writes.

The build is set up so a mistake becomes an answer an analyst corrects, not a change to your data.

Access it needs

A read-only database user that sees only the tables you allow. It can't change or delete anything.

Where data lives

Queries run inside your database. Only table names and results go to the model, in your cloud region.

Shows its working

Every answer comes with the query it ran, so nobody has to take a number on trust.

Who reviews

An analyst checks a sample of answers each week and signs off any new definition.

[07] ProofFounding client offer

Proof

Founding clientsBe one of the first two teams to run this.

The first two pilots get the AI blueprint (US$1,800) free, in return for a case study with your name on it.

Ask about it
[08] VariationsThe same build for other teams
[09] QuestionsBefore you call

Questions

Can it change our data?

No. It uses a read-only user that sees only the tables you allow.

How do we know an answer is right?

Every answer shows the query it ran and the tables it used. An analyst checks a sample each week, and wrong answers are fixed at the definition.

Which databases does it work with?

Postgres, MySQL, SQL Server, Snowflake and BigQuery, and most CRMs with an API.

What does it cost to run?

You pay the model provider per question, and your database runs the queries as usual. The pilot report shows the cost per month.

What about personal data?

Tables with personal data can be left out or masked, and people only see answers from data they're allowed to see.

Can it make charts?

Yes. It picks a simple chart when one helps, and anyone can ask for the table instead.

[10] Next30 minutes · no slides

Get this scopedin 30 minutes.

Bring the 20 questions your team asks most if you can. You leave with a written next step and a price range for the pilot.

NS
Scoping call30 min · video call

India time · confirmed on Cal ID