Use Grava
with your agent.
Connect Claude Code, Cursor, or Codex to your datasets. Grava keeps the data up to date. Your agent puts it to work.
Your agent, connected to your data. Create a dataset, read its rows, and request updates.
Connect your agent. Choose your setup guide. Keep working in the agent you already use.
Follow the guide to connect your Grava account through the API. Your agent runs where you already use it.
Direct connections for the Claude and ChatGPT chat apps are not available yet.
Create a dataset. Give your agent a public page and the columns you need. It can create the dataset in Grava and start the first collection.
“Create a dataset from this careers page. Include the role, location, and link to each job.”
Collected rows appear in your Grava dataset.
Read your rows. Ask your agent to compare prices, group open roles, or draft a report from your dataset. Check the sources behind the values it uses.
- Read the Ramp careers page. The listing shows $120–140k.
- Compared it with Monday’s record of $110–130k and marked the row as changed.
- Passed the check: 12 of 12 pages returned a value.
- Written to the record with its source.
Request an update. Ask your agent to start a collection and check the result. If an update fails a check, Grava keeps your last checked data in place.
Grava did not estimate the count from a third-party site. The cell stays empty until a source you trust publishes the value.
- This cell
- Blank
- Other cells in the row
- Preserved
- Update
- Held for your review
Connect your agent to Grava.
Choose a setup guide and start with a dataset you already use, or create your first one.