A complete guide to the Intercom agent (Fin) in 2025

If you work in customer support, you’ve probably noticed AI is everywhere these days. It seems like every team is looking for a way to manage ticket floods and just work smarter. In that conversation, Intercom’s AI agent, Fin, always seems to pop up. It’s a powerful tool that’s baked right into their platform, but is it actually the right move for your team?

This guide is a straight-up, honest look at the Intercom agent. We’re going to get into its features, unpack that slightly confusing pricing model, and talk about some real limitations you should be aware of before you sign anything. We’ll also look at why a more flexible (and budget-friendly) approach might be a better fit, especially if your team’s knowledge isn’t all sitting in one perfect, tidy folder.

What is the Intercom agent?

First things first: when you hear "Intercom AI agent," people are usually talking about one thing: Fin.

Fin is Intercom’s main AI product. Think of it as an AI-powered teammate designed to work within the Intercom platform and other helpdesks to handle customer conversations automatically. It’s built to answer complex questions, perform tasks for a user, and solve problems across your chat, email, and social media channels.

A screenshot showing the Intercom agent chatbot interface, which is the main subject of this guide.

The team at Intercom likes to say Fin is your first line of customer support, which is a good way to think about it. It’s set up to tackle three main kinds of questions:

1. The simple stuff: These are your general informational questions where the answer is the same for every customer.

2. The personal questions: Queries that need the bot to look up specific information for that user, like the classic "Where’s my order?"

3. The "do something for me" requests: This is when a customer needs the AI to take action, like processing a refund or updating their account details.

How the Intercom agent works: Setup and core features

On paper, getting Fin started seems pretty simple. But its features are so tightly wound into the Intercom ecosystem that it can create some serious headaches down the line.

Getting started with the Intercom agent

One thing you’ll hear a lot is that the initial setup for Fin is easy. You don’t need to be a developer to switch it on, which is great.

The tricky part, though, is the training. Fin learns almost exclusively from your Intercom Articles. This is the most important thing to grasp. For Fin to be even remotely useful, your entire knowledge base has to be living inside Intercom’s help center product.

A view of the Intercom knowledge base, where the Intercom agent Fin is trained.

For most companies, this is an immediate problem. Let’s be real, whose internal knowledge is that organized? If your how-to guides, troubleshooting steps, and policy documents are scattered across Google Docs, Confluence, or Notion (like they are for most of us), you’re suddenly facing a massive content migration project just to get the AI off the ground. That’s a non-starter for a lot of teams.

Key features of the Intercom agent

Assuming you’ve managed to corral all your knowledge into Intercom, Fin does have a respectable set of features:

• Works across channels: Fin can jump into conversations on live chat, email, SMS, and WhatsApp, which helps create a consistent experience.

• Customizable personality: You can tweak Fin’s tone of voice to make sure it sounds like your brand and not a generic robot.

• Can take action: Through a feature called "Fin Tasks," you can set it up to handle simple, personalized actions like checking on an order or processing a refund.

• Reporting and analytics: Intercom gives you dashboards to see what Fin is up to, how it’s performing, and where it might be getting stuck.

The customizable personality settings for the Intercom agent, allowing users to define its tone of voice.

The problem with its closed-off system

Here’s the thing: Fin’s greatest strength is also its biggest weakness. It’s so deeply integrated with Intercom that it really only works well if your entire support world lives inside that one platform.

This is where a different philosophy on AI starts to make a lot more sense. Instead of making you move mountains of documents, tools like eesel AI are built to connect to your existing mess. You can link your Intercom helpdesk, then plug in your knowledge from Confluence, Google Docs, and even past support tickets. You can be up and running in a few minutes, not a few months, and skip the painful migration project altogether.

Understanding the Intercom agent pricing model

While Fin’s features look good on the surface, the pricing is often where teams pump the brakes. It’s not just about the cost, it’s about how they charge you.

The "per-resolution" cost

The big number Intercom advertises is $0.99 per resolution. Sounds straightforward, right? But let’s dig into what that actually means. Intercom counts a conversation as "resolved" if one of two things happens:

1. A customer clicks "yes" when asked if their issue is solved.

2. A customer just stops responding after the AI’s last message.

This model creates a couple of weird issues. First, your costs go up as your AI gets better. The more tickets you successfully automate, the bigger your bill gets. It’s a strange model that punishes you for efficiency. Second, that definition of "resolution" can be a bit loose. As some users have noted, you might get charged just because a frustrated customer gave up and closed the chat, not because their problem was actually fixed.

Oh, and if you want to use Fin with a different helpdesk like Zendesk or Salesforce, you have to pay for a minimum of 50 resolutions a month.

Other costs you can’t forget

That $0.99 isn’t the only thing you’ll be paying. To even use Fin, you need a paid Intercom plan with at least one human agent seat. So your total cost is the per-resolution fee stacked on top of your regular subscription.

Here’s a quick breakdown of the plans:

A more predictable way to pay

This per-resolution model leads to unpredictable costs, which is a nightmare for anyone trying to manage a budget. In contrast, platforms like eesel AI have simple, transparent subscription plans. You pay a flat monthly fee based on how many AI interactions you expect to have, and that’s it. No surprise bills. Your costs are predictable, even when you have a super busy month.

With eesel AI’s pricing, you also get all their main products, like the AI Agent, Copilot, and Triage, rolled into one plan. It’s just more value without the weird fees and costs that scale in a way that feels punitive.

Key limitations of the Intercom agent (and when to find an alternative)

Fin is a decent tool, but its drawbacks are pretty significant. If any of these next points feel a little too familiar, it might be a sign to start looking elsewhere.

1. Your knowledge is locked in

This is the biggest roadblock for most teams. The Intercom agent is designed to work with Intercom Articles, and that’s pretty much it.

As one user on Reddit said, this 'wasn't realistic for us' because their company's knowledge was a 'mess of Google Docs, Confluence, and Notion.'

Reddit

This puts you in a tough position. You either have to dedicate weeks (or months) to a massive content migration project, or you have to accept that your AI will only know a fraction of what it needs to, which will lead to bad answers and frustrated customers.

A more flexible tool like eesel AI was designed for this exact mess. It brings all your scattered knowledge together instantly by connecting to the places your team is already working. You don’t have to change a single thing about your current tools or workflows.

2. Unpredictable and punishing pricing

The $0.99-per-resolution model is more than just an expense; it’s a business risk. If your company is growing, your support costs become a volatile expense that’s almost impossible to forecast. You get punished for successfully automating more of your support. It just doesn’t feel right.

The benefit of predictable, transparent pricing is huge. With eesel AI’s subscription plans, you know exactly what your bill will be every single month. This lets you budget with confidence and removes the penalty for scaling your support automation.

3. No way to test safely before launch

Launching an AI agent can be a little nerve-wracking. You want to be completely sure it’s going to give good answers before it ever talks to a real customer. Intercom doesn’t really offer a clear, risk-free way to test how Fin will perform on your actual customer questions before you set it live.

This is another spot where a dedicated AI platform really pulls ahead. eesel AI has a powerful simulation mode that lets you test your AI on thousands of your past tickets in a safe environment. You get a clear forecast of what your resolution rate will be and can tweak the AI’s performance before a single customer ever interacts with it. That gives you the confidence to launch without crossing your fingers and hoping for the best.

This video from Intercom introduces Fin, the AI agent for customer service, explaining its core capabilities.

The Intercom agent: A powerful tool, but not for everyone

So, what’s the final word on the Intercom agent? Fin is definitely a capable AI tool. It’s easy to switch on, especially for teams who are already living and breathing inside the Intercom ecosystem.

But the trade-offs are hard to ignore. The fact that it only works with Intercom Articles and the unpredictable pricing model create some serious hurdles for a lot of businesses.

If your knowledge is spread out across different platforms and you want a predictable bill that doesn’t go up just because you’re growing, then a more agile and open solution is probably a much better fit. You need an AI agent that works with your tools, not one that makes you work for it.

Ready to try an AI agent that adapts to your workflow? Give eesel AI a try for free and see how you can get your support automated in minutes.