Connected systems
An agent that can only talk is a phone tree. These are what let it book the meeting, update the record and open the ticket before the call ends.

An agent that cannot touch your systems can only ever promise that somebody will. The caller hears yes, a task lands in a queue, and the actual booking happens hours later if it happens at all. Reading and writing during the call is what removes that second step.
Availability, order status, account details. The agent looks it up mid-call, so what it tells the caller is true at the moment it says it.
Bookings, updates, tickets and notes go back to the system while the caller is still on the line.
You connect each system once with OAuth. No credentials are stored in your workflow, and every agent on the account reuses the connection.
A lookup that finds nothing is a path you can draw, so a missing record is a handover rather than an awkward pause.
CRMs, calendars and support tools through the integrations marketplace. What is available to your account depends on which connections you have authorised.
No. You authorise each connection once with OAuth, and the workflow references the connection rather than holding any credentials.
That is a branch you draw. The agent can retry, fall back to taking the details manually, or transfer, instead of leaving the caller in silence.
Both. Reading only would leave you reconciling afterwards, which is the problem connecting the systems was meant to solve.
Everything the platform does is available over the API, so a system with its own API can be called from a flow.
Still stuck? Talk to our team.