Agents

An Agent is an AI that works toward a goal on its own instead of following a fixed sequence of steps. You give it a system prompt, a model, and optionally tools and memory. When you run it with a task, it plans its next action, calls a tool if it needs to, reads the result, and decides whether to continue or stop — repeating that loop until it has an answer or hits a limit.

Agents vs. Workflows

	Workflow	Agent
Steps	You design the exact sequence of nodes	The agent decides what to do next at each step
Best for	Processes where the steps are known in advance	Tasks where the right steps depend on what happens along the way
Started by	Webhooks, schedules, connected apps, manual runs	A task or question, from the agent editor or its run page

A Workflow is deterministic — the same trigger always runs the same nodes in the same order. An agent is given a goal and works out the path itself, choosing from the tools you’ve allowed it to use.

Agents aren’t limited to the Agents section — add an Agent step to a workflow to call one of your agents as part of a larger automation, with its prompt mapped from earlier step data.

Opening Agents

Open Agents from the sidebar (/agents). The page lists every agent in your account as a grid of cards.

Each card shows:

Name and description
An Active status badge
A colored badge naming the AI provider the agent’s model uses (OpenAI, Claude, Gemini, …)
When the agent last ran — Just now, 12m ago, 3h ago, 5d ago, a full date once it’s more than a week old, or Never run
Edit and Delete icons, shown on hover

Click a card, or its edit icon, to open it in the agent editor.

Header: stats, new agent, sorting

Above the grid:

Total — how many agents you have
Active — how many of them are active
New Agent — creates a blank agent and opens it in the editor immediately
A sort dropdown: Newest (default), Oldest, Name A-Z, Name Z-A, Most Runs

Pagination

Choose 8, 16 (default), 32, or 64 agents per page. The footer shows the current range, for example “Showing 1-16 of 42 agents”, with page number navigation.

Deleting an agent

The trash icon opens a confirmation dialog — Delete Agent: “This action cannot be undone. Are you sure? All agent data, runs, and configurations will be permanently removed.” Confirm with Delete to permanently remove the agent, its runs, and its configuration.

Next Steps

Building Agents — create an agent and configure its model, memory, tools, and output parser on the canvas.
Running Agents — start a run, read the live run monitor, respond to approval or input requests, and cancel a run.