How to use Intercom Fin to surface relevant articles before a chat starts

Let's be honest, your support team is probably fielding the same handful of questions every single day. "How do I reset my password?" "Where's my invoice?" "What's your refund policy?" They're all perfectly fine questions, but they're also ones that could be solved in seconds with a link to the right help article.

Instead of letting these simple queries clog up your team's inbox, what if you could point customers to the answer before they even finish typing their question? That's what chat deflection is all about. By getting the right help content in front of customers proactively, you give them a way to solve their own problems. This frees up your team to handle the more complex issues that actually need a human touch.

This guide will show you exactly how to use Intercom Fin to surface relevant articles before a chat starts. It’s a great first step toward saving your team a ton of time, lowering support costs, and giving your customers the quick answers they want.

An illustration showing the Intercom Fin logo, relevant to a guide on how to use Intercom Fin to surface relevant articles before chat starts.

What you'll need to get started

Before we jump in, let's do a quick check to make sure you have everything you need. The process isn't too complicated, but it does rely on having the right setup in Intercom.

Here’s a quick list:

• An active Intercom subscription. You'll need access to features like Fin and Workflows. A quick look at Intercom's pricing page shows that Fin is on all plans, but the Workflows builder is usually on their Advanced or Expert plans.

• Admin access to your Intercom workspace. To build and edit workflows, you'll need the proper permissions.

• A solid Intercom Help Center. Your AI is only as good as the information it can find. Make sure you have clear, well-written articles that cover your customers' most common questions.

• An idea of your most common support questions. You can't deflect questions you don't know about. Take a look at your support reports to find the top reasons people are reaching out.

A step-by-step guide to setting up Intercom Fin

Alright, let's get this set up. We'll break it down into a few straightforward steps to get your bot suggesting articles.

1. Figure out what your customers are asking

Before you start building anything, take a minute to plan. The goal here isn't to automate every conversation, but to handle the easy, repetitive ones. Start by looking at your support ticket data. What are the top 5-10 questions that pop up constantly?

These are usually things like:

• Password resets

• Billing and invoice info

• Refund policies

• How to use basic features

Once you've got your list, you know exactly which help articles need to be polished and ready to go. This simple bit of analysis is your roadmap and makes sure you're automating the tasks that will actually make a difference.

2. Prepare your help articles

Think of your AI agent as a new team member who learns by reading. If your help content is confusing, out-of-date, or full of jargon, your AI is going to be just as confused.

Based on some of Intercom's own best practices, here are a few tips to make your articles AI-friendly:

• Use clear, question-based titles. Instead of a title like "Subscription Management," try something more direct, like "How do I change my subscription plan?" This makes it much easier for the AI to match a person's question to the right article.

• Keep articles focused on one topic. Don't try to explain five different things in one massive document. Stick to one problem, one solution, per article.

• Write like a human. Ditch the internal jargon and overly technical terms your customers would never use. If you wouldn't say it in a conversation, don't write it in an article.

<protip text="Intercom has a handy feature called 'Suggestions' that helps you spot gaps in your knowledge base. You can find it under "Fin AI Agent > Train > Suggestions". It looks at conversations where Fin got stuck and suggests what content you should create or update. It's a great way to improve your AI's performance over time.">

3. Build your Intercom Fin workflow

Now it's time to build the automation that will step in when a customer starts a chat. We'll use Intercom's workflow builder for this.

1. In your Intercom workspace, go to "Fin AI Agent > Workflows" and click New workflow.

2. Choose the trigger When a customer opens a new conversation in the Messenger. This makes your workflow kick off the moment someone starts a chat.

3. You can also add audience rules here if you want to get more specific, like only running this for logged-in users. For now, we can leave it open to everyone.

4. Next, you’ll want to let Fin take the first pass. After the customer sends their initial message, add a new step and choose the action Let Fin answer. This one action is the core of the whole setup. It tells Intercom to use Fin's AI to understand the question and find the best article from your Help Center.

4. Plan for a smooth handoff to your team

This is a really important step. What happens if Fin suggests an article, but it's not what the customer needed? You need a clean handoff to a human to avoid a dead end.

In the same workflow, right after Fin suggests an article, configure what happens next:

• Ask for feedback. Add a simple step that asks, "Did that help?" with "Yes" and "No" buttons.

• If they say "Yes". Awesome! You can add an action to thank them and automatically close the conversation. One less ticket for the team.

• If they say "No". If the article wasn't helpful, or if Fin couldn't find an answer to begin with, the conversation needs to go to a person. Add an action to assign the conversation to your support team. This makes sure no one gets stuck in a loop.

Common pitfalls (and how to fix them)

Getting this workflow running in Intercom is a great start. But as you grow, you might bump into a few limitations with this kind of setup. Here are some things to think about as you move from a basic bot to a more powerful one.

The problem with a single source of knowledge

The biggest catch with this approach is that Intercom Fin can only read what's in your Intercom Help Center. But let's be real, is that where all of your company's knowledge lives? Probably not. What if the best answer is in a Google Doc your product team just updated, or a detailed guide on a Confluence page? Fin will never see it. This leads to the bot failing and your team having to answer the question anyway, even though the info was already written down somewhere.

This is why tools that connect all your knowledge sources are so important. A platform like eesel AI can plug into your helpdesk and all your other apps. By giving your AI the full picture, it can pull from everywhere to find the most accurate and helpful answer.

The hidden cost of paying per answer

Intercom's AI pricing seems simple at first, but it can get expensive. You pay your base subscription, plus $0.99 for every resolution Fin handles. While that doesn't sound like much, it means your bill goes up the more successful your bot is. If you have a busy month with lots of deflected tickets, you could get a surprisingly big bill. You're basically paying a penalty for being efficient.

In contrast, eesel AI has predictable pricing based on a set number of monthly interactions, with no extra fees per resolution. This lets you scale up your automation without worrying about costs spiraling out of control. You know what you're paying each month, no matter how many questions your AI answers.

The risk of launching blind

While you can preview your workflows in Intercom, it’s hard to know how your bot will really do against thousands of actual customer questions. You're often launching with blind spots, making it tough to guess your actual resolution rate or figure out where the AI is likely to get tripped up.

A big advantage of a more advanced tool like eesel AI is its simulation mode. You can test your AI agent on thousands of your past support tickets in a safe environment. This gives you a really accurate forecast of its performance, shows you exactly how it will respond to different questions, and helps you calculate your cost savings before it ever talks to a real customer.

A good start on your automation journey

Using Intercom Fin to surface relevant articles before a chat starts is a smart way to handle common questions and make your support team more efficient. By following the steps we outlined, you can build a solid foundation for customer self-service.

But as we've seen, this setup has its limits, especially around knowledge sources, predictable pricing, and testing. It’s a great place to start, but if you're serious about building a smart, scalable automation strategy, you'll eventually want a more flexible solution.

The smarter way to automate support

If you're ready to move past the limitations of a siloed, unpredictable AI, you should take a look at eesel AI. It's designed to work with the tools you already use, not make you start over from scratch.

• Go live in minutes. Connect your helpdesk and all your knowledge sources with one-click integrations. No complex setup or developers needed.

• Connect all your knowledge. Don't limit your AI to a single help center. eesel AI learns from your Google Docs, Confluence, past tickets, and more to give customers the best answer every time.

• Test without the risk. Use powerful simulations to get your setup just right and launch knowing exactly what to expect.

• Pay one flat, predictable price. Our plans are simple and transparent, with no per-resolution fees to worry about.

Ready for an AI support solution that actually works with your team? Try eesel AI for free and see how quickly you can automate your support with confidence.