Short answer: n8n and Make both host AI agents that can answer in Slack, and they fit best when your team already runs workflows there. Runbear fits when the goal is one shared agent the whole team @mentions in Slack or Microsoft Teams. Runbear hosts the agent for you, so an ops manager can add an AI agent to Slack with no engineering team and no server to host.
Disclosure: Runbear is our product. n8n calls its agents a preview and Make calls AI Agent (New) an open beta, so details may change.
Hosting, setup and pricing at a glance
| n8n Agents | Make AI Agent (New) | Runbear | |
| Where it runs | n8n Cloud, or the self-hosted Community Edition with extra setup | Make's platform, in the data zone your organization picks (US1, US2, EU1 or EU2), though some AI features are processed outside it | Hosted by Runbear |
| How it reaches Slack | Slack is one of the agent's channels, and n8n is a launch platform for Slack's Add to Slack | A scenario with a Slack Watch New Events module and a Send a Message module | A custom bot with its own name (needs a Slack configuration token you generate in Slack), or the shared @Runbear bot |
| Setup | Describe the agent, pick a model, attach tools and workflows | Build a scenario with a Slack trigger, the agent, a reply module and a filter | Describe the job, connect knowledge and tools, pick the channel |
| Pricing model | Monthly workflow executions. One agent turn is one execution. Every plan includes unlimited users | Make credits: 1 per operation, plus credits for AI tokens when you use Make's AI provider | Charges by usage in credits, not by seat |
| Tool permissions | One credential per tool. A tool marked sensitive waits for approval | The connection you pick on each module | Per-User or Shared, where the tool supports both |
| Knowledge | Uploaded csv, pdf, markdown or txt files (n8n Cloud), or a vector store | Knowledge files the agent keeps, such as FAQs and company policies | Notion, Google Drive, Confluence, SharePoint, Slack, a website or local files |
n8n Agents: best when your team already builds in n8n
An n8n agent has instructions, a model and tools. The tools can be MCP servers, built-in n8n integrations or your existing workflows. The agent decides when a step runs, and the workflow fixes what that step does. The same agent can answer in Slack, run on a schedule, or be called from another workflow. According to n8n, "You don't need to know how workflows work to build one." Instead, you describe the job in everyday words.
You choose where it runs. n8n Cloud hosts the agent for you. The Community Edition is a self-hosted version on GitHub, and agents run there with some extra setup. Self-hosting is the only setup in this comparison where you keep a server running yourself. If you write the Slack bot yourself instead, a serverless platform such as Vercel runs the code, but the code and each deployment stay with you. n8n still labels its agents a preview, and it recommends testing an agent before you publish it and requiring approval for sensitive actions.
n8n is one of the 10 launch platforms for Slack's Add to Slack. In Slack's announcement, n8n product manager Sindhuja Jeyabal said: "Building and publishing an AI agent in n8n is easy, though getting the whole team to actually use it can be challenging. The Add to Slack is what makes it easy with just a couple of clicks."
Make AI Agent: best inside existing Make scenarios
Make AI Agent (New) is the newer agent app Make released in February 2026. Its agents live inside scenarios, Make's name for automated workflows, and Make's docs say some familiarity with Make helps. A Slack agent is a scenario that watches a channel for new messages and replies with a Send a Message module. Make's guide adds one step you should not skip: put text such as "Agent:" before each reply and filter on it, or the agent answers its own messages in an infinite loop. The guide also mentions an optional Slack app with a bot connection to tell agent and user messages apart, which takes more setup.
Make's own advice fits all three tools. Use an agent for tasks that need judgment and have variable inputs, and keep a plain scenario for work that should run the same way every time. Check one thing before you start: Make is not in the Slack Marketplace, so a workspace that allows only Marketplace apps needs an admin to change that setting.
Zapier Agents: a short note
Zapier Agents work across the 9,000+ apps Zapier supports, and the template gallery includes an IT Helpdesk Slack Responder. Zapier's Slack app can also start a Zap when a new message is posted in a channel. Zapier's pricing page counts agent work in activities, with 400 a month on the free Agents plan, and lists agent sharing across an organization on Agents Enterprise. Our Lindy vs Zapier vs Runbear comparison covers Zapier in more depth.
Runbear: best when the whole team shares one agent in Slack
Runbear hosts the agent, so there is no server to keep alive. You describe what the agent should handle, connect the documents it answers from and the tools it can act in, and add it to Slack or Microsoft Teams. Runbear also keeps the knowledge sources synced.
- Where a tool supports both options, you set it to Per-User, where every member links their own account and the agent acts with that member's permissions, or to Shared, where one account serves everyone.
- The agent always answers @mentions and DMs. In each channel you can also turn on replies to keywords or to every new message.
- A trigger runs the agent on a schedule, when a webhook arrives, or on an event in a connected app.
- Agents can act in 2,000+ tools. On the Claude Agent SDK runtime, the usual default, you choose Claude, GPT or Gemini for each agent.
Two limits to plan for. A tool connected as Shared gives everyone who uses the agent that account's access. Agents pause when your organization reaches its credit limit, unless an admin allows overage on a paid plan.
On pricing, the Team plan is $99 a month with 20,000 credits and 5 builders, the members who create and manage agents.
When to choose Runbear, and when not to
Choose n8n if you want to self-host, already run n8n workflows, or have someone technical who will own the agent. Choose Make if your processes already live in Make scenarios and the agent is one step among many. Choose Zapier Agents if your team already runs Zaps.
Choose Runbear if you want a named agent that a whole team uses in Slack or Teams, that answers from synced docs, and that acts in tools with each person's own access where the tool supports it. If your team does not work in Slack or Teams, one of the workflow tools will fit better.
FAQ
Can I use n8n or Make together with Runbear?
Yes. A Runbear HTTP / Webhook trigger gives you a URL that any system able to call a webhook can send a request to, so an n8n or Make workflow can start a Runbear agent. The call doesn't wait for the agent's reply, so you get the result from the trigger's alerts.
Do I need to register my own Slack app?
Not to get started, though it depends on the tool.
- n8n: Slack says n8n agents connect through its Add to Slack flow.
- Make: the standard setup uses Make's own Slack app, and a separate Slack app with a bot connection is optional.
- Runbear: the shared @Runbear bot installs from the Slack Marketplace, so you approve one app instead of registering your own bot. A custom Runbear bot with its own name needs a Slack configuration token, which you generate in Slack.
Sources
- Introducing n8n Agents (n8n blog) and n8n pricing
- Anyone can now build Agents on Slack: Introducing Add to Slack (Slack blog)
- Make Help Center: Introduction to Make AI Agent (New), Slack-triggered AI agent, Credit usage for AI agents, AI providers and data zones in Make and the Slack app documentation
- Vercel
- Zapier Agents, Zapier pricing and How to get started with Slack on Zapier
- Runbear: How to add an AI agent to a Slack channel, Slack integration, pricing, and the docs on tools, chatbot settings, triggers and billing
Vendor pages were checked in October 2026.
