A practical guide to your Intercom chatbot setup in 2025

Thinking about setting up an automated chatbot in Intercom? Good call. You’re probably hoping to spend less time on repetitive questions, free up your support team for the tricky stuff, and get answers to your customers faster. If that sounds about right, you’re in the right place.

You have two main paths you can take. You can stick with Intercom’s own AI agent, Fin, which is built right into the platform. Or, if you want a bit more muscle and control, you can bring in a specialized AI tool. This guide will walk you through the native Intercom chatbot setup step-by-step. But we’ll also be real about the common snags you might hit and show you a more flexible alternative for teams who want more control and a better bang for their buck.

What you'll need for your Intercom chatbot setup

Before you jump in, let’s quickly run through what you’ll need. It’s a short list, but having these things ready will make the whole process go a lot smoother.

• An active Intercom account with admin permissions. Keep in mind you’ll need a paid plan to use the Fin AI Agent.

• Access to your website’s code or a tool like Google Tag Manager if you haven’t installed the Intercom Messenger yet.

• A collection of your existing support content. This can be your help center, an FAQ page, or any public documents you use to help customers.

The native Intercom chatbot setup (using Fin)

For many teams, starting with Intercom's Fin just makes sense. It's already there, which makes it feel like the easiest option. Let's walk through getting it up and running.

Step 1: Train Fin with your support content

An AI chatbot is only as helpful as the information you give it. To get going, you’ll need to point Fin toward your existing support content.

You can find this in your Intercom workspace by going to "Fin AI Agent > Train > Content". From there, you have a couple of ways to feed it information:

• Intercom Articles: If you're already using Intercom’s Help Center, this is the simplest route. Just pick which articles you want Fin to learn from, and it will pull information straight from your knowledge base.

• External URLs: You can also add links to any public web pages. This is handy if your guides, policies, or FAQs are on your main website. Just paste the URLs, and Fin will crawl them for information.

A screenshot showing how to connect various knowledge sources for an Intercom chatbot setup.

This is a decent starting point, but you might spot a big limitation pretty quickly. Most of a company's real knowledge, the stuff that actually solves tough customer problems, isn't sitting on public web pages. It’s usually tucked away in internal tools like Confluence, Google Docs, or buried in the details of thousands of past support tickets. Fin can't get to any of that, which often leads to knowledge gaps and a lot of "I don't know" answers. An alternative like eesel AI connects instantly and securely to all those scattered sources, giving the AI the full context it needs.

Step 2: Configure basic bot behavior with simple deploy

Once Fin has some content to work with, you can get it live quickly using the "Simple deploy" option. It's designed to get you started without getting lost in the weeds of complex settings. You'll find it under "Fin AI Agent > Deploy > Chat".

Here are the main settings you’ll want to look at:

• Audience: Choose who gets to talk to the bot. You can make it available to everyone, just new visitors, logged-in users, or even specific groups of customers.

• Introduction Message: Write the welcome message Fin will use to greet people. This is your chance to set the right tone.

• Handover: What should happen when Fin doesn’t know the answer? You need a fallback plan. Usually, this means passing the chat to a human agent, but you could also point them to other resources.

• Auto-close: Decide if and when conversations should be closed automatically after a customer stops responding. This helps keep your support inbox from getting cluttered.

An image displaying the chatbot guidance and behavior configuration page in the Intercom chatbot setup.

Step 3: Create custom logic with workflows

If the simple setup feels a little too restrictive, you can dig deeper with Intercom Workflows. This is how you build out a more advanced Intercom chatbot setup by creating custom rules for how chats are handled.

You can get to this from the "Advanced setup through Workflows" section. Here’s a quick example of a common scenario: routing conversations based on what a customer writes.

1. Create a new workflow that kicks off "When a customer sends their first message."

2. Add a "Let Fin answer" step. This puts the AI on the front line, giving it a chance to solve the issue before it reaches your team.

3. Next, add a "Branches" step. This lets you send the conversation down different paths. For instance, you could set a condition for one branch: "if Message content contains "billing"".

4. If that condition is met, you can add an action that assigns the chat to your finance team. All other chats can be routed to your general support inbox.

This image shows the workflow builder, a key part of an advanced Intercom chatbot setup.

Workflows are pretty useful, but they can become a real headache to manage as you add more and more rules. For teams that want that same fine-tuned control without building complicated flowcharts, eesel AI provides a fully customizable workflow engine that’s much simpler to use. You can set up precise automation rules, from simple keyword routing to complex actions, all in a more intuitive way.

Step 4: Test and launch your chatbot

Whatever you do, don't just switch on a new bot for all of your customers at once. Testing is an essential step to make sure things are working right and to avoid creating a frustrating experience for everyone.

Intercom has a preview tool that lets you chat with your bot just like a customer would. You can ask it some common questions and see how it responds based on the content and rules you've set up.

A view of the testing environment for an Intercom chatbot setup, where you can preview its responses.

Once the preview looks good, we suggest rolling it out slowly. Start by turning the bot on only for your own team or for a small segment of customers. This lets you see how it performs in the real world while the stakes are low, so you can fix any issues that pop up.

A live preview is a great sanity check, but it won’t tell you how your bot will handle a high volume of chats or what your actual resolution rate will be. This is where a tool like eesel AI's simulation mode comes in handy. It lets you test your entire AI setup on thousands of your actual past support tickets in a safe environment. You get a solid forecast of resolution rates, cost savings, and quality before a single customer ever interacts with it.

Common pitfalls of the native Intercom chatbot setup (and how to solve them)

While Intercom's Fin can handle some basic automation, many teams quickly bump into a few common problems. Here’s what to look out for and how you can get ahead of them.

Pitfall 1: Unpredictable and scaling costs

One of the biggest surprises for teams using Fin is the bill they get at the end of the month. The pricing is based on a $0.99 fee for every resolution the bot handles.

The issue here is that your costs become totally unpredictable. In a weird way, you get penalized for being successful. If you have a busy month and your bot resolves a ton of tickets, you’re looking at a huge, unexpected bill. This makes budgeting nearly impossible and can make you hesitant to automate more of your support.

Solution: Look for transparent, predictable pricing. For example, eesel AI's plans use a flat monthly rate that includes a large number of AI interactions, with no extra fees per resolution. You know exactly what you’re paying each month, which lets you scale up your automation without worrying about costs spiraling out of control.

Pitfall 2: Limited knowledge sources and context

We touched on this earlier, but it’s worth repeating. The native Intercom bot is cut off from almost all of your company's most useful knowledge. It can't learn from past ticket resolutions in Intercom, internal wikis in Confluence, or standard operating procedures in Google Docs.

The result? The bot gives vague answers, or worse, just gives up and escalates tickets that it could have easily solved with the right information. This kind of defeats the whole point of automation and leaves both customers and agents feeling frustrated.

This diagram contrasts limited versus extensive AI knowledge sources, a key consideration for your Intercom chatbot setup.

Solution: Great AI support comes from unified knowledge. eesel AI was built specifically for this. With dozens of one-click integrations, it securely learns from every single place your team keeps information, including your historical Intercom tickets. This gives it the deep context needed to provide accurate, helpful answers that reflect your company's voice and solutions.

Pitfall 3: The risk of a 'rip and replace' migration

When a team outgrows what Intercom's native bot can do, they often think the only option is to move to a completely different helpdesk platform just for better AI.

That’s a massive, painful project. A full migration can take months of work, cost a fortune in developer time, and force your entire support team to learn a new tool from the ground up. It’s a huge disruption for a problem that can be solved much more easily.

Solution: Instead of ripping everything out, just plug in an upgrade. eesel AI is a "plug-and-play" solution that integrates directly into your existing Intercom inbox in minutes. You get a much more powerful and intelligent AI agent without having to leave the tools and workflows your team already uses every day.

Elevate your support with the right Intercom chatbot setup

Getting started with the native Intercom chatbot setup is a totally reasonable first step into support automation. It can handle simple questions and give you a taste of what’s possible. But it does come with real limitations around cost, knowledge sources, and customization that can hold you back as you grow.

The good news is you don’t have to accept those trade-offs or plan a massive migration to a new platform. A smart AI strategy is about finding a tool that works with your current systems, gives you full control, and offers a clear, predictable return on your investment.

Ready for a smarter Intercom chatbot setup?

Upgrade your support with an AI that plugs directly into Intercom, learns from all your team's knowledge, and comes with simple, predictable pricing. Try eesel AI for free and see how much you can automate in just a few minutes.