How to deploy an AI customer support agent using Intercom

A quick guide to launching an AI support bot in customer service solution Intercom

Tags Streamline Icon:
Wrench Streamline Icon:
People Man Graduate Streamline Icon:

In this tutorial, you’ll learn how to use Intercom to deploy an AI agent and automate customer support workflows.

AI agents provide instant and accurate answers to customers directly in chat interfaces. These tools act as the first line of support triage, directly interfacing with customers to answer inquiries, before looping in human agents.

These workflows are not currently possible with ChatGPT or custom GPTs yet, so we’ll be using Intercom, a leading AI-first customer support platform, for this tutorial.

If you use another customer support platform or are currently evaluating options, don’t worry -  this tutorial will still have practical applications for whichever AI-powered customer support platform you choose.

In the first section of this tutorial, we’ll provide an overview of some of the tools in the market to help provide a map of the territory.

You’ll need:


  1. A sampling of AI agent tools in the market
  2. Develop a customer support AI agent
  3. Ticket classification and automation

A sampling of AI agent tools in the market

We’ll be working with Intercom in this tutorial, but as previously mentioned, this isn’t the only option in the market. There are several tools available, which can be bucketed into two categories.

AI-Powered Customer Support Platforms

These are more robust suites of tools that can do everything from deploying AI agents, hosting your internal and external knowledge hub, routing customer inquiries, providing deep analytics, and more. The more prominent providers in this space include:

AI Chatbot Builders

Newer entrants in the space have started as chatbot-only offerings. These are not singularly-purpose-driven customer support tools but are highly configurable AI chatbots that can be embedded on existing websites or standalone pages and used as customer support AI tools.

💡 Tip: We are not recommending one of these solutions over the other; it's highly dependent on your business needs and budget. This non-exhaustive list is intended to help you get started in your evaluation.

Develop a customer support AI agent

To start developing a customer support AI agent in Intercom, you’ll want to upload content for the bot (aka ‘Fin’ in Intercom) to reference. This can be plain text snippers, PDF uploads, or public URL sources. You can also point the Intercom AI Agent to your Intercom Help Center content if you already use that feature within Intercom.

To add content, click on the Fin AI Agent button on the left side navigation and then the New Content button in the top right corner.

In this example, I’ve added Brand Guidelines, FAQs, and an Escalation Procedures document.

In addition to content, you can add Custom Answers that act as exact responses for specific questions. Custom Answers allow you to create bespoke answers to your most important questions, and the AI agent will prioritize them over its AI Answers.

To add Custom Answers, navigate to Custom Answers in the left side navigation, click on New Answer in the top right corner, and complete the form.

💡 Tip: Make sure to set the Custom Answer live to ensure your AI agent uses it.

Once you’ve uploaded your content and added Custom Answers, you can start testing your AI agent in the Overview window. Click on the Intercom logo in the bottom right corner on the Overview page and ask it a question from your documentation.

💡 Tip: It’s prudent to test not only known questions from your documentation and custom answers, but also unknown questions and duplicitous requests, to get a solid understanding of how your AI agent performs across a series of queries.

Once you’re satisfied with the quality of your tests, you can finish the setup to set it live. This includes defining the audience that will encounter your AI agent and routing mechanisms based on issue resolution status.

To set up these features, go to the Overview page and click on the “Set up and go live” tab. You can set up custom routing logic and audience groups in this tab.

For this example, we’ll set the audience to Users, Leads, and Visitors. For routing logic, if our AI agent cannot resolve the conversation, we’ll assign the ticket to our Support team.

💡 Tip: You’ll need to install your AI agent on your website or app as well using Javascript or no code implementations as part of the setup process. Detailed instructions on how to do this are provided on the Intercom Overview page.

Ticket classification and automation

In addition to AI agents, Intercom provides a suite of workflow tools that will enable you to effectively route conversations to the right teams, tag conversations with specific keywords, and triage support threads in an automated fashion.

To get started, navigate to the Automation page from the left side navbar, select the Workflows sub-page, and click New Workflow.

In this example, we’re going with the “Route conversations to the right team” template. Below is the base template for this workflow.

We’re going to update this template so the AI agent we developed in the previous step will answer the customer query first. If there is no resolution, we assign it to a team based on whether the word “billing” is mentioned or not.

To do this, click “Add Step” and select “Let Fin answer”, which deletes the branches step. Then add back the Branches step, build a branch for if “Message content contains billing”, and re-point this branch's step to the existing Assign paths steps.

Now, our AI agent will triage customer questions first, and route them to specific teams automatically based on the branch logic we’ve defined.

💡 Tip: There are many workflows you can explore in the Intercom workflow builder. We suggest starting with their templates and customizing them as a first step.

This tutorial was created by Garrett.

Join 5,163 others learning. Create a free account

Sign up with
Already have an account?
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

Get full access

✔️ All 100+ courses & tutorials in our catalog
✔️ New content added weekly
✔️ Private community access
✔️ No subscription, $250 paid once
✔️ Expense it using this template. Or get a team account.
✔️ 30-day refund policy. No questions asked
Join 5,163 learners from companies like Microsoft, Coca Cola, NBA, Adobe & Google

More tutorials like this

View all

If you scrolled this far, you must be a little interested...

Start learning ->

Join 5,163 professionals already learning