Watch our Fin AI Copilot launch event

Senior Managing Editor, Intercom

April 11, 2024

Intercom continues to lead the way in building the future of support with our AI-first Customer Service platform.

Following up on the groundbreaking release of our Fin AI Agent last year, we have now launched Fin AI Copilot, which gives every customer service agent their own personal AI assistant. Fin can find information, help solve problems, and complete all of those time-intensive tasks in seconds. And you can trust the answers Fin provides are accurate because it pulls from your support materials and even conversation history.

We’re building a world where every agent becomes a superagent with an AI Copilot by their side. An AI Copilot that’s connected to the entire platform, is constantly learning, and can proactively problem solve alongside your human agents in real time.

Check out the FIn AI Copilot launch event, featuring:

What follows is a lightly edited transcript of the announcement broadcast.

Welcome to Built For You

Eoghan McCabe: Hey, thank you for joining this broadcast. We’re calling it Built For You, and this is going to be the first in a series of monthly broadcasts just like this.

The idea here is to be more open and more vocal about the technology that we’re working super hard to bring you before anyone else. Each month you’ll get to hear directly from the people who built the tech, and today we have Olivia and Brian from our product org to walk you through a major new release that we’re really excited about, a new AI product release. It’s probably our biggest in a year.

About 18 months ago, we decided to focus our efforts entirely on customer service. We want to be the next major platform and we sincerely believe we have a shot. There has been very little movement in this space for a while. Just one month after that, the world changed – OpenAI released ChatGPT, and we quickly realized that we must react or die. Believe it or not, we decided to react and we pretty quickly found ourselves betting the company on the simple idea that AI is going to flip the nature of customer service on its head.

“We were the first to build an AI customer service agent”

We wanted to lead that charge and compete using our pace of innovation. We were the first to build an AI customer service agent. We called ours Fin. We shipped it to beta in March of last year and a whole year later and counting, we’re still ahead of the incumbents – none of them have released an AI agent of their own.

Today is our next major release of three core components that make up the Intercom AI customer service platform, which we’ll tell you a little bit about later. These components will represent a fundamentally new way of solving the customer service problem in a way that’s cheaper, faster, and better. It’ll please both customer service leaders and their customers. So without further ado, let’s jump in. We have a little video to try get you excited about our announcement, and then the announcement itself will follow. Thank you. We hope you’ll enjoy.

Fin AI Copilot in action

Olivia Singarella: Hi, my name’s Olivia and I’m on the product marketing team here at Intercom. Today I am super excited to introduce our latest AI innovation, Fin AI Copilot.

When we spoke to customer service agents, we consistently heard the same thing. They’re spending way too much time searching for answers to customer questions. They’re searching across multiple tabs and various sources that take them away from the customer. It’s inefficient and keeps agents from focusing on what really matters, building deeper lasting customer relationships.

Fin AI Copilot is here to help. Fin is a personal AI assistant for every agent and will dramatically improve your team’s efficiency. Agents can access Fin directly in the inbox and get instant expert answers from multiple sources, including your team’s conversation history, internal and external content.

“Fin instantly understands the context of the conversation and has already suggested a question that the agent can ask”

Let’s see Fin in action. Here, we have a customer conversation happening in the Intercom inbox, and you can see Fin is already ready to help in the right-hand panel. Looks like the customer needs help with a refund. Fin instantly understands the context of the conversation and has already suggested a question that the agent can ask.

When you ask Fin a question, it scans across your entire knowledge base and generates an instant answer that is personalized to the current customer conversation. Let’s take a deeper look. Fin connects to external sources like public health center articles, internal sources like agent handbooks and third-party platforms like Confluence, Notion or Guru. But what really sets Fin apart is that it’s the only AI Copilot that can source answers using your team’s previous conversation history.

Here, Fin has provided a customer ready reply that includes information from the top two most relevant sources. If you hover over the first source, you can see this was pulled from a public article. You can easily click into the article and view the entire source without leaving the conversation. Fin will even highlight the exact passage it pulled from, so you can easily validate.

If you hover over the second source, you can see it’s pulled from an agent’s past conversation. Similarly, you can click and see exactly which answer it pulled from, from the context of the conversation so you can verify. In addition to Fin’s AI-generated answers, you can also view all of the other sources that Fin found when scanning your knowledge base. So now agents can get both AI answers and browse relevant sources in one central place in the inbox. When you’re ready to send, simply click “Add to the composer”, and if you want to personalize the reply, you can use AI to instantly translate, adjust to your tone, make it more friendly or formal right from the composer. Let’s go ahead and update to your tone of voice with just one click and send it off to the customer.

“Fin immediately gets to work against synthesizing all of the most relevant sources to answer this question”

Now, what happens if the customer has a follow-up question? It looks like they’re hoping to get an exception for their refund. Let’s follow up with Fin and ask what if the order was over 60 days ago? Fin is always contextually aware, so if you need to ask a follow-up question, Fin will provide an answer based on the previous question asked. Fin immediately gets to work against synthesizing all of the most relevant sources to answer this question.

This time, Fin has generated an answer using an internal article that provides the agent’s guidance on how to proceed next. As always, you can easily hover and click into the source to view and validate. And because this answer was generated using internal articles, we’ve included a reminder to check that the content is shareable before sending off to the customer. When the answer’s all set, I can send the next response back to the customer again with just one click.

You’ve now seen how Fin AI Copilot works and it’s already proving to help agents dramatically increase their efficiency. At Lightspeed agents using Fin AI Copilot, we’re able to close 31% more customer conversations daily compared to agents not using Fin. And at Deliverect, Fin has helped their team increase their response speed by up to 75%, enabling them to provide personalized high-quality and fast response times that customers expect these days. The wait list is now open and you’ll find more details about how to sign up at the end of the broadcast.

AI-first is the future

Brian Donohue: So we think you’re going to love AI Copilot. It’s going to fundamentally change how your team works inside the inbox. But Copilot is one part of a broader system we’re building. We now have a new and better way to deliver excellent customer service and it’s AI first. This is both a product change and a mindset change. Let me explain in more detail how we’ve been building out this vision.

So we’re building a single system that delivers AI first experiences for customers, for agents, and for CS leaders, all built on the foundation of Intercom’s complete service platform. So it starts with customers. Our AI agent provides instant accurate answers to most customers’ questions. Of course, it’s available 24/7 so people don’t have to wait. Only the more complex questions get passed through to human agents. You know that part, right? But this is not the old world of clunky, naive keyword-based chatboys that customers just have to put up with.

“We have hundreds of customers above a 70% resolution rate”

This is the new world of fluent, smart AI agents built on the latest generations of tech, which are capable of delivering human-like quality of service. So as you may know, we launched our Fin AI agent last summer. We were pretty much first to offer this new gen of tech, but those speed of market was important. Our top priority was making it as accurate as possible by preventing hallucinations when it makes stuff up, we wanted to stop that. So you have an AI agent that you can actually trust. And since our launch, the performance of Fin, it just keeps getting better and better. We take a huge amount of pride in this. Our average resolution rate across all our active customers, it’s now approaching 45%, and we have hundreds of customers above a 70% resolution rate.

Those kind of numbers show you just how much Fin can transform your support operation, and this is actually the first part of the mindset change – every support team needs to be using an AI agent.

So we’re going to keep making Fin better and better, like giving you controls for Fin’s tone of voice, making Fin work over email. Fin’s going to get smarter at troubleshooting and Fin’s going to be working with data. In fact, we already have an alpha where Fin intelligently connects to third-party systems. Fin’s actually figuring out should I go to this system and how to do it and what to pull out? And it’s giving real-time dynamic answers to things like your eCommerce inventory. What’s new here is that Fin is doing all the heavy lifting to make this happen. So that resolution rate, that’s just going to continue to climb.

“We are building these AI solutions together within a single system, so customers get compounding value”

Now for the questions that our AI agent just can’t resolve, they get rooted to your team. And of course that’s now where AI Copilot comes in. This is the next pillar of our system.

Olivia just showed you how AI Copilot works, but a key point to understand is that our AI agent and AI Copilot, it’s not an either-or it’s not an “oh, if you don’t want to use the AI agent, well then you should consider a Copilot.” We are building these AI solutions together within a single system, so customers get compounding value from using both. What’s really exciting is that we’re now seeing in the beta the AI Copilot as actually most effective when your team is using AI agent to handle those easier questions. And that’s because AI Copilot is built to help your teams with the hardest ones. That’s where the big time savings come into play, and that’s how these products give your team compound value. And this is the second part of that mindset change. Your team will need to break their habits of how they do customer support.

Now, AI Copilot will be their first stop and often their last. Now the third pillar of our AI-first platform is AI analyst. Service teams care most about the quality of the customer experience, but the tools to measure that today, they give you a partial blurry view of what’s actually going on, and then even when you can kind of see what’s going on, it’s time-consuming and actually frequently impossible to figure out why. Why is happening and what should I do to make it better?

“Last year, we shipped 230 product improvements to our platform”

So we’re building out AI insights and growing that into a fuller AI analyst solution to solve those problems. We’ve already released some of the first steps here, like our new unresolved question reports that uses AI to highlight what are the questions you need to actually answer with better content. And this speaks to the third part of the mindset change, using AI to help you figure out how to optimize your support operation, and usually that’s going to mean improving your content, and that’s going to mean you need to spend a lot more of your team and your time on creating better content for AI.

Now, this AI system is only possible because we’re building it on, and integrating it deeply into, Intercom’s complete customer service platform. And just the last year alone, we made our world-class platform even better with major updates to tickets, to phone, to email, reporting and Workflows, which is our inbound orchestration tool. Last year, we shipped 230 product improvements to our platform. Oh, we’re just going to keep at this, continuing to make those products more robust, more comprehensive with an even higher product quality.

We are determined to make Intercom the best and the only customer service platform you need. So go get yourself on that AI Copilot wait list. Thanks for listening. Thanks to all our customers who’ve been on this journey with us. AI is just getting started. So are we.