Agent update
Agents builder
n8n Agents, launched on 25 Sep 2026, are standalone agents you set up once with instructions, a model and tools, then chat with, drop into a workflow, connect to Slack, or run on a schedule. The same published agent is used in every one of those places.
How to use n8n Agents for small business
A support lead could describe a ticket-triage agent, attach workflows that look up account context and add a CRM note, and let the team talk to it in Slack, each in their own conversation. A sales coordinator could put a research agent on inbound leads and have it post a short summary to Slack, with approval required before it writes to the CRM. An office manager could schedule a morning agent that checks what changed overnight and sends a brief, or call the same agent from a ticket-intake workflow.
Agents are on n8n Cloud for everyone on the latest stable version, on community self-hosted with some extra setup, and in preview for Enterprise. n8n says they are still in preview, so test before publishing and keep approvals on anything that writes to a system of record. One turn counts as one execution, and tool calls to workflows or sub-agents do not count separately. They share your workflow execution quota.
Most useful for: Automation Consultant – Build and test workflows and AI agents in tools such as Zapier, Make or n8n, Marketing Operations Manager – Build and manage marketing automation and lead routing, Office Manager – Coordinate workflow and deadlines across the team
Frequently asked questions
What does it cost, and is it still in preview?
Yes, n8n says agents are still in preview. One turn counts as one execution, tool calls to workflows and sub-agents do not count separately, and agents share your workflow execution quota. Building an agent with n8n Assistant uses AI credits. You can try a model with Gateway credits without bringing your own API key first.
How is this different from the AI Agent node I already use in workflows?
The AI Agent node is unchanged, and existing builds keep working. The new Agents are a standalone type with memory, sessions, channels, versions and approvals included, and you can still use workflows as tightly scoped tools. When a workflow needs an agent for one step, you add the Message an Agent node and call the same published agent.
Once I publish an agent, where can the team use it?
The same published agent can be used in every connected place: you can chat with it, talk to it in Slack or other channels, run it on a schedule, or call it from a workflow with the Message an Agent node. Change the instructions once, publish, and every place it is connected gets the new version.