Ask a question. Get the answer and the math behind it.
It knows how your team defines each metric, so revenue means your revenue. Every answer comes with the chart, the SQL it ran and a check against the source before it replies.
Searched catalog · orders, customers
Ran 2 queries · Snowflake
Nine accounts. Valley Fresh Markets is the biggest gap: $42,100 in this month last year and nothing yet. Their last order was 38 days ago.
| Customer | Owner | Last order | Same month LY |
|---|---|---|---|
| Valley Fresh Markets | J. Ortiz | 38 days | $42,100 |
| Bluebird Co-op | M. Hale | 41 days | $9,600 |
Queries used
select customer, owner, last_order_days, gross_ly from gold.monthly_orders where ordered_every_month_ly and mtd_gross = 0
Checked: MTD gross matches the orders table, $1.06M
It learns how your business counts before it answers
Each org gets its own semantic layer: what the tables mean, how they join and how your team defines each metric. When the leasing team says net rentals, the analyst uses the same formula the dashboard uses.
Tell it something new about the business and it saves it as context after you confirm, so the next question starts from there.
Metric definitions
Cedar Point Residential
Net rentals
Move-ins minus move-outs for the period, by property.
yardi.unit_events
Physical occupancy
Occupied units over rentable units. Down units are excluded.
yardi.rent_roll
Total delinquency
Resident balances past the current bucket, as of the period end.
yardi.resident_aging
Every answer comes with the queries behind it
Answers arrive as charts, tables and KPI cards built from live data. Open the queries to see the SQL. For metric questions the analyst checks its result against the source totals and tells you what it checked.
Like the chart? Add it to a dashboard in one click.
- Line, bar, area, scatter, treemap and pie charts
- Tables you can sort and export to CSV
- Share a conversation with a teammate
Queried spend by platform · BigQuery
Joined first-party orders by first touch
Blended CAC rose to $38.40. Social A did most of it: $92,400 of spend for 2,210 new customers, $41.81 each on first-party attribution.
Checked: new customers match the orders table, 4,849
Pick the model. Switch on Zero Data Retention per person.
Choose from 26 models across 6 providers and set your own default.
Turn on Zero Data Retention for a user and the analyst only offers the 22 models available under zero-retention terms.
Model
26 models from 6 providers
- OpenAI
- Anthropic
- xAI Grok
- Moonshot Kimi
- Z.ai GLM
It can do the follow-up, after you say yes
Connect the apps your team works in and the analyst can act there: create a task in Asana, draft a reply in Gmail, update a record in Zoho CRM or HubSpot. It works with 26 connected apps.
Nothing is written until you approve that specific action. You see the app, the record and the fields first.
Approval needed · Asana
Create a task in Account follow-ups
- Task
- Call Valley Fresh Markets about October order
- Assignee
- J. Ortiz
- Due
- Tomorrow
Type it, say it or attach it
Voice input
File uploads
Prompt Library
Memory
In Slack
From your AI tools
AI analyst FAQ
How do I know an answer is right?
Every answer lists the queries it ran, so you can read the SQL. For metric questions the analyst also checks its result against the source totals before it replies, and tells you what it checked.
Can it change my data?
Database connections are read-only. The analyst can take actions in connected apps such as Zoho CRM or Asana, and it asks you to approve each one before it runs.
Which AI models can I use?
26 models from 6 providers: OpenAI, Anthropic, Google, xAI Grok, Moonshot Kimi and Z.ai GLM. Each person sets their own default.
What does Zero Data Retention do?
Switch it on for a user and the analyst only offers models available under zero-retention terms, 22 of the 26. Requests from that user go out with retention turned off.
What if it gets a definition wrong?
Tell it. The analyst saves business context after you confirm it, and the metric definitions for your org are where you can see and fix how each number is calculated.
The rest of the analyst
What customers say
“The best way I've explained it: it's like reading the Wall Street Journal about your specific use case.”
Chad BuckendahlChief Data & Analytics Officer, Truly Free Home“One prompt turns into a dashboard that's incredibly useful for somebody, and it took me 15 minutes.”
“I had to search through thousands of records today for something that isn't easily searchable in Zoho, and Cerebro did it very quickly and then gave me the SQL code that I could run myself to generate a report to share with the client for verification.”
Ask your first question today
Connect one source and ask what you'd normally ask an analyst. Read the SQL, check the total and decide for yourself. Free to start.