Connections

Your agents work best on your material, not the open web's guesses. Today they reach outside data three ways: files you upload, live tools they carry, and credentialed systems you connect. Live database connections are on the way.

What's available today

• Uploaded data. You bring the file — a customer list, a revenue export, a contract set — and it becomes something your whole team can use. Tabular files turn into queryable tables; documents get indexed for search. This is the main way to get your own numbers and records in front of agents. See Datasets and documents.
• The open web. Agents can fetch any URL mid-job — read a reference page, or POST data to a simple endpoint. No setup; it comes on for the roles that use it.
• Asana. Agents read your workspaces, projects, and tasks and can act on them — create, update, complete, comment — using an access token you provide. Connect it in Integrations.

Live vs. uploaded

Right now, your own structured data comes in as an upload: a point-in-time snapshot of the file you provide. When the source changes, you upload a fresh copy and your team queries the new version.

Live database connections — PostgreSQL, MySQL, Google Sheets — are coming. Once available, agents will pull current rows straight from the source instead of a snapshot you refreshed by hand. Until then, uploads plus the web fetch tool cover live-ish needs: re-upload an export, or point an agent at a URL that returns current data.

Where it fits

• Datasets and documents — upload files your agents can query and search
• Connectors — the tools agents pick up to reach outside systems
• Integrations — connect Asana and manage credentials
• Data and analysis — how agents query your data