How to set up an Intercom chatbot: A step-by-step guide

Let’s be honest, setting up an Intercom chatbot can feel a little daunting. But it’s also your ticket to getting a handle on all those repetitive questions, qualifying leads while you sleep, and giving customers the instant answers they expect. When it’s done right, your team gets to breathe and focus on the trickier problems that actually need a human touch.

This guide is here to walk you through the whole thing, no fluff. We’ll cover how to get Intercom’s own AI chatbot, Fin, up and running. We’ll look at the quick-and-easy "Simple deploy" method and then dive into the more customizable "Advanced setup through Workflows."

While Intercom’s tools are pretty solid, there’s definitely a learning curve. We’ll break it all down so you can get your bot working for you, not against you.

What you’ll need before you start

Getting a few things in order first will save you a headache later. Before you jump in, make sure you have these sorted:

• An active Intercom account: This one’s a given, but you’ll need a plan that actually includes the Fin AI Agent feature.

• The Intercom Messenger: This should already be installed on your site or in your app. It’s how your bot will chat with visitors.

• Support content: Your chatbot needs something to learn from. Go ahead and gather up your help center articles, support docs, and any other public URLs you want it to use for answers.

• Clear goals: Why are you doing this? Are you hoping to cut down on simple tickets by 25%? Book more sales demos? Knowing what you’re aiming for makes it much easier to tell if it’s actually working.

How to set up an Intercom chatbot in 4 steps

Getting your Intercom chatbot live is a process of feeding it information, deciding how it should behave, and then fine-tuning its performance. Let’s get into it.

1. Train your chatbot with the right knowledge

An AI bot is only as good as the information you give it. The quality of your knowledge base is everything, so this first step is probably the most important one.

Head over to the Fin AI Agent section in your Intercom dashboard. This is where you’ll tell the bot what to read.

• Connect your Intercom Articles: The simplest starting point is your own help center. Just enable your Intercom Articles as a knowledge source, and the bot will start pulling answers from the content you’ve already written.

• Add external content: You can also point the bot to information outside of Intercom. Add public URLs from your company blog, developer docs, or other websites. Fin will sync this content to get a broader understanding of your business.

• Provide specific guidance: This is where you give the bot a bit of personality. In the "Guidance" section, you can define its tone of voice (friendly? formal?), tell it how to handle tricky subjects, and set rules for when it needs to pass a conversation to a human. This helps the bot sound less like a generic robot and more like a part of your team.

2. Choose your setup path

Intercom offers two main ways to get your chatbot running. One is for getting started quickly, and the other is for people who love to tinker.

Option A: The simple setup

The "Simple deploy" option is exactly what it sounds like: the fastest way to go live. It’s a great fit for teams who want to start answering common questions right away without building complicated conversational paths.

The setup is pretty direct. You’ll just need to configure a few settings:

• Defining the audience: Decide who the bot should talk to. You can make it available to everyone, or limit it to just logged-in users or new website visitors.

• Selecting channels: Choose where the bot will show up. You can turn it on for your website, mobile apps, or even connect it to channels like WhatsApp and Slack.

• Customizing the handover: Set clear rules for when the bot needs to give up and find a human. This is key to making sure customers don’t get stuck in a frustrating loop.

This path is your best bet if your main goal is to deflect common questions and you don’t need a ton of custom logic.

Option B: The advanced setup using workflows

If you want fine-grained control, the "Advanced setup through Workflows" is what you’re looking for. This lets you build custom conversation flows using a visual editor. It takes more planning, but you can build a bot that handles much more than simple Q&A.

For example, you could build a workflow with different branches. The bot could ask a question, and depending on the answer, take the customer down a different path. A basic workflow might look something like this:

1. A customer starts a chat.

2. The workflow scans the message for the word "billing."

3. If it finds it, the chat automatically goes to the finance team.

4. If not, it checks for the word "demo."

5. If it finds "demo," the bot can launch a calendar app to book a meeting.

6. If neither keyword is there, Fin will try to answer the question using its knowledge base.

Workflows let you create really personalized experiences, but they can get messy to manage as you keep adding more rules.

3. Configure key chatbot behaviors

No matter which path you take, you’ll need to adjust a few settings that shape the customer experience.

• Handover Experience: Nobody likes getting stuck talking to a robot that can’t help. A smooth handoff to a human is essential. You can configure the bot to ask a few clarifying questions before escalating, which can sometimes help it find the answer or at least give your human agents more context.

• Auto-close Conversations: To prevent your team’s inbox from getting cluttered, you can set rules for when the bot should automatically close chats. For instance, if a customer gets an answer and doesn’t respond for 10 minutes, the conversation can be marked as complete.

• Conversation Ratings: How do you know if the bot is doing a good job? Turn on CSAT (Customer Satisfaction) surveys to ask for a quick rating after an interaction. This feedback is gold for figuring out what’s working and what isn’t.

4. Test and launch your chatbot

Please, please, please do not launch a chatbot without testing it. A bad bot experience is often worse than no bot at all.

• Use the preview tool: Intercom has an interactive preview where you can chat with your bot and see how it responds. Use this to check for accuracy, tone, and just general helpfulness.

• Roll out gradually: Don’t unleash your new bot on every single customer at once. A good first move is to deploy it internally so your own team can kick the tires. They’ll help you spot any weird phrasing or wrong answers before customers do.

• Analyze and optimize: Once you go live, the job isn’t over. Keep an eye on your Intercom analytics. Look at the bot’s resolution rate, its CSAT scores, and what kinds of questions it keeps failing to answer. That data tells you exactly where you need to improve your help articles or tweak your workflows.

This tutorial offers a detailed guide on how to make a custom AI chatbot for Intercom, giving you more control over your automation.

A simpler alternative to Intercom’s chatbot

While Intercom’s built-in chatbot is a decent tool, many teams hit a few common roadblocks when they try to scale up their automation.

Common limitations of Intercom’s chatbot

• Workflow Complexity: That powerful workflow builder can get out of hand. As your business adds more products or services, managing a web of rules and branches can become a full-time job.

• Siloed Knowledge: The bot mostly learns from your public help articles. Getting it to securely access information from private sources like Google Docs, Notion, or past tickets in a different helpdesk can be a real pain.

• Unpredictable Pricing: The "$0.99 per resolution" model](https://www.intercom.com/pricing) sounds great at first, but it can lead to some surprising bills. A busy month can cause your support volume, and your costs, to jump unexpectedly.

• Deployment Confidence: The preview tool is handy for checking single questions, but it doesn’t really tell you how the bot will perform in the wild. It’s tough to guess what its real-world impact will be across thousands of unique conversations.

A smarter chatbot with eesel AI

This is where a tool like eesel AI can make a huge difference. It doesn’t replace Intercom; it plugs right into it to fix these exact problems, giving you a faster and smarter way to automate support.

Getting started is surprisingly straightforward.

• Go live in minutes: eesel AI for Intercom connects to your account with a single click. You don’t need a developer or a series of sales calls. You can get a capable AI agent running on your own, sometimes in less than an hour.

• Unify your knowledge: This is a big one. eesel AI doesn’t just read your public articles. It connects to over 100 other tools, including private knowledge in places like Google Docs, Confluence, and Notion. It can even learn from your past support tickets, giving your bot a much more complete picture of your business.

eesel AI's dashboard shows how it connects to multiple knowledge sources beyond public articles, a key step when you set up an Intercom chatbot alternative.:

• Test with real data: Remember that guessing game with performance? eesel AI has a simulation mode that lets you test your bot on thousands of your actual past support tickets before you ever turn it on. It gives you a clear forecast of its resolution rate and shows you exactly where it shines and where it needs more help.

The eesel AI simulation feature provides a clear performance forecast, which is a powerful tool when you want to set up an Intercom chatbot with confidence.:

• Benefit from predictable pricing: With eesel AI’s pricing, you can forget about per-resolution fees. The plans are based on feature tiers with plenty of interactions included, so your bill won’t suddenly double just because your bot is having a successful month.

eesel AI offers transparent, tier-based pricing, a predictable alternative to Intercom's per-resolution model for those learning how to set up an Intercom chatbot.:

When you put Intercom’s native bot and the eesel AI integration side-by-side, the differences become pretty clear. With Intercom’s bot, you’re looking at hours or even days of setup time, especially if you’re building complex workflows. eesel AI gets you going in minutes with a self-serve setup.

The biggest distinction is in the knowledge sources. Intercom’s bot sticks to public articles for the most part, while eesel AI securely connects to over 100 sources, including all that useful internal knowledge in places like your past tickets, Google Docs, and Confluence. This alone leads to more accurate and helpful answers. And when it comes to testing, you can move from a simple live preview to a full-blown simulation on your historical data, giving you a much clearer picture of performance before you launch.

Pro tips for a successful chatbot setup

Whether you stick with Intercom’s native bot or bring in a tool to help, these habits will set you up for success.

• Pro Tip

  Don't set and forget: An AI bot needs regular check-ups. Carve out time to review its conversations and analytics to find and plug knowledge gaps. A tool like eesel AI can help by showing you exactly what questions are stumping your bot, so you know what content to create next.

• Start with a narrow scope: Don’t try to automate your entire support operation on day one. Pick your top 5-10 most frequent, simple questions and train the bot to handle those perfectly. Once it’s doing well, you can slowly give it more responsibility.

• Nail the handover: Make the jump from bot to human as smooth as possible. The bot should always set clear expectations, letting the customer know when an agent will be available if it needs to pass the conversation on.

Taking your Intercom chatbot to the next level

Setting up an Intercom chatbot is a solid move toward a more efficient support team and happier customers. The process is manageable: feed your bot good information, choose a setup that fits your needs, configure its behavior, and test it well before it talks to customers.

While Intercom’s own tools give you a good starting point, they do have their limits. For teams that want to get results faster, use all their company knowledge, and keep costs predictable, a solution like eesel AI for Intercom is a clear winner. It enhances your existing setup to deliver smarter automation without the usual headaches.

Ready to improve your Intercom support?

Stop wrestling with clunky workflows and surprise bills. eesel AI works with Intercom to give your customers better answers, automate more of your team’s work, and show you a clear return on your investment.

Start a free trial or book a demo to see how the simulation feature works with your own data.