Using agents
What a team member does with agents in Pi Durable Platform's console, and where people normally talk to agents instead.
Most people never open Pi Durable Platform's console. They talk to agents in their team's own app, such as a chat app, and the app reaches the agent through the platform's API. The console is where a team's members test agents, look at their conversations and manage how apps reach them.
As a member of a team you can:
- see the team's agents in Agents, and every agent of every team you're on in All agents;
- chat with an agent on its Test tab, in shared or private threads (Testing an agent, Shared and private threads);
- deploy and use agents of your own in your personal workspace, if your platform has them (Your personal workspace);
- create API keys for your own scripts (API keys);
- connect apps that act as you, and disconnect them (Connected apps).
What you can't see
Some agents in a team are service-managed: an app created them for its own users, one per user. Only the team's service accounts talk to them, so you don't see them as a member. Team owners and admins see that they exist (with a Service-managed badge) and can delete them, but can't read them. See Deploying agents.
Another person's private threads are invisible to you, as if they didn't exist.
Agents that are asleep
An idle agent hibernates: its runtime stops, its conversations stay. The next message wakes it, which takes a few seconds. The agent's Overview says whether it is running now. You don't need to do anything to wake it.