Automation agents

Agents that run multi-step workflows across the tools a team already uses.

13 agents, each with its inputs, outputs and review level checked by a person.

Automation agents run multi-step work across the tools a team already uses: they watch for a trigger, decide what to do, call the right apps and hand back a result or a record. The difference from a fixed workflow is judgement in the middle: the agent reads the email, the ticket or the document and chooses the steps, within the limits you set. The listings record what each agent connects to, how it is triggered, what it returns, and whether a person approves actions before they run.

Choose by how much of the process is predictable. If every case follows the same path, a classic workflow tool is cheaper and easier to test. If cases vary and need reading and deciding, an agent earns its place. Check the setup level: several of these products are platforms that need configuration or a developer, and their pricing is often per task or per run.

The agents

Questions

How is an automation agent different from a workflow builder?

A workflow runs fixed steps. An agent reads the input and decides which steps to take, calling tools as needed. Several products here are workflow platforms that added agent steps; their listings say which parts are agentic.

Can I require approval before the agent acts?

Many products offer approval steps, and the review level on each listing reflects the vendor's documented default. Where actions run without review the listing says so.

What do automation agents cost?

Per task, per run or per seat, often with a free tier. The listing shows the vendor's published model and starting price with the date it was checked.