Analyze and query your data

This guide shows you how to ask the Obvious agent questions about your sheet data — filtering, grouping, summarizing — and save the results as new sheets you can keep working with.

Ask a question, get an answer

The agent can read every sheet in your project. To analyze your data, describe what you want to know in the chat panel. No special syntax, no formulas. Just ask.

Which customers have revenue over $100k?

Show me a breakdown of deals by region

What's the average order value per product category?

List everyone whose contract expires in the next 30 days

The agent reads your sheets, runs the analysis, and responds with results right in the chat. If the answer is a handful of rows, it shows them inline. If it's a larger dataset, it creates a new sheet so you can browse and filter the results yourself.

You can ask follow-ups too — the agent remembers context:

Now filter that to just the West region

Group those by account owner instead

How it works under the hood

The agent translates your question into a structured query and runs it against your sheets. Your original data stays untouched — every query reads without modifying anything.

Save results as a new sheet

When an analysis produces something worth keeping, ask the agent to save it:

Save that as a new sheet called "High-Value Customers"

The agent creates a new sheet with the filtered or summarized data. You can specify a destination too — "Save those results in my Q4 Analysis workbook" — and the agent adds a sheet there. That saved sheet is a real artifact you can build on: add columns, set up validations, create views, or use it as input for another analysis.

Combine data across sheets

The agent can pull from multiple sheets in a single question.

Compare revenue in the Deals sheet with targets in the Goals sheet

Show me customers from Accounts who don't have any entries in the Activity Log

The agent matches records across sheets based on shared fields — customer name, email, ID, whatever connects them. It's the equivalent of a VLOOKUP or a database JOIN, except you describe the relationship in plain language and the agent handles the rest.

Get more specific

The more specific your question, the more useful the answer:

Filtering: "Show me deals over $50k that are still in negotiation"

Grouping: "Break down support tickets by priority and show the count for each"

Calculations: "What's the total pipeline value, grouped by stage?"

Combine them freely. "Top 5 regions by total revenue, but only for deals closed this quarter" works just as well as a single filter.

For developers: SQL is available

The agent also accepts SQL directly. Reference your sheets by name:

SELECT "region", SUM("revenue") as total_revenue
FROM "Deals"
WHERE "status" = 'Closed Won'
GROUP BY "region"
ORDER BY total_revenue DESC

Type SQL in the chat, or ask the agent to "run this SQL against my Deals sheet" and paste it in. The full DuckDB dialect is supported — window functions, CTEs, subqueries, JOINs across sheets. Results can be saved as new sheets the same way.

Tip: Ask the agent "what columns are in my Deals sheet?" to see available fields before writing a query.

Next steps

Validate and clean your data — Set up rules to catch errors and flag records that need attention.

Create views from your sheets — Turn query results into Kanban boards, calendars, or timelines.

Working with the Agent — Get more out of the agent with better prompts and advanced workflows.