Monday questions, Monday answers
Your data has answers. Your BI seat count is zero.
Enterprise BI sells six-figure contracts to companies with analysts. oat.ai serves everyone else: the ops lead with a Postgres login and a question.
The old way
- Ask the one person who knows SQL. Wait.
- Or buy BI: $100K+ contracts, months of setupTypical enterprise BI deployment range — Vendr / Mammoth 2026 pricing guides
- Dashboards answer last quarter's questions
- Chat bolt-ons hide the query they ran
With oat.ai
- Type the question. Get chart, table, answer.
- $0 to start with a CSV. $29 for a warehouse.
- Every new question answered in seconds
- SQL shown on every answer — audit it yourself
What you get this week
From CSV to answered question before Friday.
Inside your first seven days
- A workspace and your first CSV answered in minutes
- 20 free questions — chart, table, and SQL on each
- Shareable answer links your team can open, no login
- On Starter: one live warehouse connection, read-only
- On Starter: /oat questions answered inside Slack
What it costs
- Free: CSV mode, 20 questions a month — forever
- Starter: $29 per month, 500 questions, 1 connection
- No contract, no seat minimums, no sales call
- Cancel anytime; nothing of yours is stored
Accuracy that compounds
It learns what "active user" means at your company.
Generic chatbots guess at your schema. oat.ai keeps a semantic layer per workspace — your definitions, your metrics — so every correction makes the next answer better.
The SQL is always shown
Every answer carries the query that produced it. Auditable, copyable, correctable. Chat bolt-ons that hide their SQL ask for blind trust; we never do.
rows of your data stored. Queries run read-only and proxied; results stay yours.
A memory per workspace
Metric definitions, naming quirks, fiscal calendars — remembered. Accuracy compounds per customer, not per model release.
Answers where you work
/oat revenue last quarter by plan — full answers inside Slack, plus scheduled Monday digests in English.
Scored in the open
We benchmark our own text-to-SQL accuracy and publish the scorecard. Measured honesty, not marketing claims.
Pricing
The mid-market entrance the giants can't afford to build.
Databricks and Snowflake answer questions inside their own walls. ThoughtSpot starts where budgets end. We start at zero.
Free
- CSV & Sheets mode
- 20 questions / month
- Shareable answer links
Starter
- 1 warehouse connection
- 500 questions
- Slack app
Team
- 3 connections
- Semantic layer
- Monday digests
Business
- Unlimited connections
- Anomaly alerts
- Embed widget + SSO
Asked before connecting
The questions that come before the questions.
What happens to my data?
Nothing is stored. Connections are read-only, credentials live in a vault, and queries are proxied — results render and stay yours. Export everything, any time.
What if it gets an answer wrong?
You'll see it: the SQL is on every answer. One click flags it, your correction updates the workspace's semantic layer, and that mistake doesn't repeat. Phrasing is confidence-gated — it tells you which tables it used.
We're already on Snowflake. Why not Cortex Analyst?
If all your data and all your askers live inside Snowflake, use it. oat.ai is for the rest: plain Postgres, mixed sources, CSVs, and teammates who will never open a warehouse console.
Is this a replacement for our BI tool?
For most small teams it replaces the need for one. Dashboards answer the questions you predicted; oat.ai answers the one you just thought of.
Who is behind oat.ai?
A small team of data practitioners who wanted a tool their non-analyst colleagues could actually use. We run our own metrics through it every Monday.
Your next question takes nine words. Not nine months.
Upload a CSV and ask. Twenty questions a month, free, no card — $29 a month when you connect a warehouse. The SQL comes with every answer.
Ask my first question — free