AI digital worker · Data

Illustrated persona of the AI data worker
05

Luke

Data analyst

Answers your data questions. Shows its work, every time.

Luke answers the questions your team asks every day, straight from your warehouse, spreadsheets and documents. Every answer comes with the query and sources behind it, so anyone can check the work.

Modes
CopilotAutopilot
Reports to

Your data lead

Works in
SnowflakeBigQuerySlack
Illustrated persona of the AI data worker

Question in #revenue

Illustrative example
DoneAutopilot

Head of sales

How did net revenue by region change last quarter?

Illustrated persona of the AI data worker

Asia Pacific grew the most, up 17.0% on Q2. North America rose 5.4%. Europe dipped 2.1%.

SQL

WITH q AS ( SELECT region, SUM(CASE WHEN fiscal_quarter = '2026-Q2' THEN net_revenue END) AS q2, SUM(CASE WHEN fiscal_quarter = '2026-Q3' THEN net_revenue END) AS q3 FROM finance.revenue_by_day WHERE fiscal_quarter IN ('2026-Q2', '2026-Q3') GROUP BY region ) SELECT region, q2, q3, ROUND(100.0 * (q3 - q2) / q2, 1) AS change_pct FROM q ORDER BY q3 DESC;

Result

RegionQ2Q3Change
RegionNorth America
Q24.10M
Q34.32M
Change+5.4%
RegionEurope
Q22.85M
Q32.79M
Change-2.1%
RegionAsia Pacific
Q21.12M
Q31.31M
Change+17.0%
Source

finance.revenue_by_day (Snowflake)

Definition

Net revenue, as approved by your finance team

Names, companies and figures are fictional.

See Luke at work

Names and glyphs are personas for software agents. They help your team know who handles what.

Illustrated persona of the AI data worker

The job

The job description

Six tasks Luke takes on, inside the systems your team already uses.

01

Answer plain-language questions

Takes business questions the way your team asks them, in plain language.

02

Show the SQL

Writes the query, runs it, and shows it with every answer.

03

Build charts and tables

Posts them where your team already works.

04

Send scheduled reports

Delivers reports to Slack, Microsoft Teams or email on your schedule.

05

Watch your metrics

Keeps an eye on the metrics you choose and flags unusual changes.

06

Combine your sources

Brings warehouse data, spreadsheets and documents together in one answer.

Two ways to work

Luke drafts or finishes. You set the line.

We recommend starting every task in copilot and moving it to autopilot once your team is happy with the results.

Copilot

Luke prepares. A person approves.

Illustrated persona of the AI data workerA question for leadership
Waiting for approval

Asked why churn rose last month, Luke returns a chart, the SQL behind it and the caveats it found.

An analyst checks the work before it goes to leadership.

Analyst can: approve · edit · send back

Autopilot

Luke finishes the job, inside your guardrails.

Illustrated persona of the AI data workerThe Monday revenue summary
Done

Every Monday morning, Luke posts the weekly revenue summary to your leadership channel.

It uses the same approved definitions each time.

You choose the mode for each task, and can change it anytime.

CopilotAutopilot

A week on the job

A week with Luke

MonWedThu

Monday

Posts the weekly metrics summary to the leadership channel.

Autopilot

Any day

Answers ad-hoc questions in Slack with a chart and the SQL behind it.

Autopilot

Wednesday

Answers a question on churn with a chart, the SQL and caveats, for an analyst to check.

Copilot

Thursday

Flags an unusual change in a metric you asked it to watch.

Autopilot

An illustrative week. The real one depends on your tools, your volume and the rules you set.

Luke prepares. A person approves.Illustrative example, not a customer result.

Guardrails

What Luke won't do without you

Luke follows these rules on every task. Add your own on top, and every action is logged.

How oversight works
01

Read-only by default

Connects with read-only access by default. Nothing in your warehouse is changed.

02

Every number is checkable

Shows the SQL and the sources behind every number it reports.

03

Your metric definitions

Uses the metric definitions your team approved, and asks when a question is ambiguous.

Tools

Where Luke works

Luke works inside these systems, with the access you grant. Your team keeps its tools, and the work shows up where it always has.

Each connection is set up with you during onboarding, and availability varies by system, so we confirm yours before you commit.

Product names are trademarks of their respective owners.

SnowflakeBigQueryDatabricksPostgreSQLAmazon RedshiftGoogle SheetsSlack

FAQ

Before you hire Luke

Still have a question? Talk to us
You need someone who owns your metric definitions, not a full data team. That person approves the definitions Luke uses and reviews its answers while tasks run in copilot mode. If you already have a data team, Luke takes on the routine questions and scheduled reports, so your analysts can spend more time on deeper work.
Yes, every one. Each answer comes with the SQL Luke ran and the sources it used, so anyone on your team can rerun the query or question the logic. Luke uses the metric definitions your team approved, and when a question could mean two different things, it asks before answering.
Not by default. Luke connects with read-only access, so nothing in your warehouse is changed. It reaches only the data sources you grant, and every query it runs is logged. Any broader access would be a separate, deliberate decision by your team, and you can revoke access from your side at any time.
Yes. Luke can combine warehouse data with spreadsheets and documents in one answer. For example, it can compare actuals from your warehouse with the targets your team keeps in Google Sheets, and point to the document a definition came from. Each answer lists the sources it used, so you can see where every number came from.

The team

Meet the rest of the team

Luke covers data questions. The rest of the team covers sales, support, finance and HR, with the same approvals and audit trail.

01Sales
Illustrated persona of the AI sales worker

Vanessa

Sales development rep

Researches accounts, writes outreach worth answering, and books meetings for reps.

SalesforceHubSpotGmail
Meet Vanessa
02Support
Illustrated persona of the AI customer support worker

Caleb

Customer support agent

Answers customers in their language and hands tricky cases to people.

ZendeskIntercomWhatsApp Business
Meet Caleb
03Finance
Illustrated persona of the AI finance worker

Hannah

Finance operations specialist

Handles invoices, collections and reconciliation. Payments always wait for your approval.

NetSuiteQuickBooks OnlineXero
Meet Hannah
04HR
Illustrated persona of the AI HR worker

Spencer

People operations coordinator

Screens applicants, schedules interviews and gets new hires ready for day one.

GreenhouseWorkdayGoogle Calendar
Meet Spencer

Ask Luke the question you've been waiting on.

Bring a question your team asks every week. We'll show you how Luke would answer it from your data, with the SQL behind it.

Early access is open. No commitment.