Your first analysis
Get an answer from the shared web corpus, then take it into your own workflow.
Start with what’s already there
You don’t need to collect pages before querying. Your organization can analyze the shared corpus, including captures collected by others.
- Open SQL workspace and paste the query below.
- Run the query. Each returned row is a capture: an observation of a page at a point in time.
- Open Tables to inspect the available columns, or Functions to explore search and extraction helpers.
SQL
SELECT capture_id, url, observed_at
FROM captures
ORDER BY observed_at DESC
LIMIT 20;Build on a useful result
Use the Python SDK to continue in a notebook or application, or connect an AI agent over MCP. Save a query when you want a reusable starting point—not a stored result.
| If you need… | Use… |
|---|---|
| A specific page that isn’t covered yet | A capture to fetch one known URL now |
| Pages you can reach from a starting point | A crawl that follows links from your URLs |
| The same source material for future questions | A pin to keep exact captures longer |
| Observations to compare over time | A schedule to capture the same pages repeatedly |
If the result is empty
An empty result is not a failed query. There may be no matching data in the current corpus. Try a broader query, inspect the schema, or capture or crawl the sources you need. Collection and availability in SQL are separate steps; a completed capture may not be queryable immediately.