Name: Introduction to Intercom & Fin AI Agent
Uploaded: 2026-04-30T16:16:09.401Z
Duration: 41 min 32 s
Description: Introduction to Intercom & Fin AI Agent

Transcript for "Introduction to Intercom & Fin AI Agent": Hello. Hello, everyone. Hello? Hi, guys. I'm Elena from the product specialist team here of Intercom and Fin. Yeah. So we're hosting this webinar together with Kendall. Hello? Yeah. Great to have you all here. We are really excited to give you an introduction to Intercom and FIM and work you through what FIM, our AI agent, can potentially do. We probably gonna give everyone two minutes to join to make sure that everyone who wanted to join us here and that you're all ready to go. In the meantime, maybe you can share where are you joining from. We have a chat on the right side. Yes. As you see, Kendall is from Chicago. I am from Dublin. Anyone else is willing to share? Malaysia, Boston with you guys? India? Cool. Hello? Orlando. K. We have a lot of people. Spain. Nice. China. K. That's great. And, yeah, while we are waiting for people, we want to share a very short poll on, where are you on your journey with Intercom and Fin. If you're currently on a free trial with us, if you are already a customer, or maybe you haven't started yet and you are just exploring different options of different help desks and AI agents, We are going to share the poll on your right side as well next to chat. So please, feel free to vote. It's gonna help us a lot. So, ma'am, so we'll do it a couple more seconds before we get started. Yeah? Yep. Perfect. Yeah. So just to set up the expectations, yes, by the end of this webinar, we hope that you will get a very good understanding of how, Intercom and Thin is helping support teams to deliver fast and smarter customer service at scale and how you can actually use it for your support environment. So, yeah, I think, two minutes passed, so we are ready to start. Yes. Here is a quick look at the agenda we have for today. We will start by giving you a very short introduction to Intercom, and then Kendall will jump and show you a live demo of our Intercom suite. After that, we will take a closer look at our AI agent, Finn. We'll start with a bit of a background on how we got there and why AI agents are making such a big difference right now in the support space. Then Kendall will jump in and show you a demo of Finn. At the end, we will look at some of our customer success stories, and then we will have some time for Q and A as well. You can write your questions in the Q and A section. It's also on your right side from the screen. We will take care of them if you have any questions while we're making this presentation. Let's get started. Just a bit of a refresher for those who may not be familiar with Intercom. The Intercom customer service suite combines the number one, AI agent for customer service with the next generation help desk. And, it is built, on a single platform that maximize team efficiency and delivers superior service. And, we are the only one complete AI first customer service platform on the market at the moment. So what does it mean? We've made the strategic decision to be AI first in everything we do. And, our solution combines industry leading AI capabilities for everyone, for your customers, for your support agents, and for support leaders as well, with a complete omnichannel support platform. We founded Intercom in, 2011, and we pioneered live chat for customer service. We've been on the edge of CS innovation ever since. Now let's zoom into the Intercom Helpdesk. Our Helpdesk is fast, modern, and it is built to scale with you and with your team, and it is powered by AI to maximize agent productivity. So our Copilot acts like a personal assistant for your team, helping agents respond, faster, stay on track, and also work more efficiently with your customers. Intercom help desk is designed for speed and collaboration, giving your team everything they need in just one place from tickets and macros to some real time context automation and reporting. With the reporting tools, you can continuously monitor the performance and optimize the customer experience to drive even better results. And finally, we offer more than, four fifty integrations across a wide range of different apps and services alongside with a flexible API. So you can actually easily, connect Intercom Helpdesk to your existing tech stack. You will see the Helpdesk connection now during the demo, but the key takeaway here is that everything is designed to reduce friction, both for your teams and for your customers as well. Now let's jump into the fun part. Yes. So demo, Kendall will show you a demo of Intercom Helpdesk. In the meantime, I'm gonna close the old poll, and I want to ask you how you're doing with your current support setup. I'm gonna share another poll. So while Kendall is sharing his screen, you can maybe vote. Thank you. Fantastic. Thank you, Elena. Hello, everyone. I'm Kendall, and I'm gonna give you a quick breakdown of the intercom help desk. There are four quadrants in our help desk for omnichannel support, and we're gonna go from left to right. This first section is where your inbox and all of the things that you can find, tickets, conversations, etcetera. Here, you can see your open conversations. Here is your apply box, and then here is more of, like, the back end data. So starting from left to right, you have your inbox. This is where any conversation that you have assigned to you will live. You have your mentions or anything that you're mentioned in, anything that's created by you, your all tab for all the conversations in your workspace, anything that's unassigned. You have your spam tab and then the dashboard. Now we have a couple different options for Fin. We have Fin for service. Fin for sales currently set up in our workspace. So So Fin for service, any service conversations, anything that's resolved, escalated, and handed off, anything that's pending, any spam for Fin, then Fin for sales, all of your Fin for sales specific conversations, things that are routed to your team because they're disqualified, pending, and escalated to support. As we scroll down, we can see our team inboxes, and these are completely customizable. Currently, we have a general support inbox, a premier support team, and a sales team. You can pin these all to this section. Here, you have your teammate section. You can look at your teammate's inboxes if you wanna see any conversations that they're working on. We have views as well. Views are customizable, so if you wanna see specific tickets, and the ticket type is billing. Or any Messenger conversations with the current channel is your Web Messenger, or the current channel is Android, iOS, these are all completely customizable as well. Moving over to the conversations, this is where all of your conversations will live. We have a couple different states for conversations and tickets because in Intercom, we have both. You have open, closed, and snoozed states for conversations, and then your ticket statuses, so submitted, in progress, waiting for customer, and resolved. You can adjust this by activity, the date started, the next SLA if you have SLAs in your plan, and then you can change this from ascending to descending order. Here's my name, and here's the search bar. If I wanna search any conversations that I have, if I have, it's kinda open. Moving over to the second or the third quadrant, the, reply box. Here, we have our customer's name. You can mark conversations as priority. You can ask Fin for Copilot. When you click the ellipses, you have a couple different options as well. You can manage participants, you can merge conversations, create a new conversation, export the conversation to text, PDF, or show conversation events. The conversation events is really cool because if you click on this, it shows you everything that's going on in the conversation specifically. So anything that's triggering in the background, any workflows that are running, etcetera. Now you have the option to convert to a customer ticket because as I said, we have conversations and tickets on intercom. Conversations are a very quick back and forth, and tickets, if you need to dive deeper into things and probably investigate things, you can use a ticket for that. We have intercom phone button if you would like to make a call from the inbox. You can snooze your conversations, and we have a couple of different states here, from tomorrow, a week, or a custom snooze state. And then you have the option to close a conversation. Now moving on down here in the reply box, this is where the entire conversation takes place, and we can reply to conversations here. You can send it directly from. Now you have the option to reply or write a note. So for instance, say that I need a little help from Melina, I can at mention Melina and send a note in the conversation. Now this is completely internal for your team, so your customer does not see the note section. Now a couple of different options as well in the reply box. You can pin things by clicking the ellipses. And if you wanna create macros, emojis, GIFs, you wanna send articles from your help center, any attachments, images, tours, or you can use AI compose feature where the AI can summarize a conversation. You can change the tone of voice, make it more friendly, more concise, more formal, etcetera. Now moving on to this last quadrant, we have our details tab and Copilot. Now the details is typically things that your customer don't see, but it's all going on in the background. So any conversation that's assigned to me is the assignee, and the team inbox, you can add it to the team inboxes like we mentioned before. We have a couple different ticket options. So we have customer tickets, tracker tickets, and back office tickets. All of these will be found here in the links. And then side conversations, if you wanna have a side conversation with your teammate, on the side end of the conversation. Moving on down to conversation attributes, this is all the detail within the conversation, and this is also completely customizable as well. So you can see an ID, a brand, a subject. You click on see all. You have a couple of different customizable options as well that you can add to your conversation. You can also create custom conversation attributes. So if you want to track specific data points and customize them, you can add that to your conversation attributes as well. We have a topics button, which will formulate topics that are genuinely popular with the conversations. And as they grow, you get that reporting data. You have your user data. This is everything that you need to know about your customer. So their name, the company they work for, their user or lead. You can also completely customize this with custom data attributes as well. So as we see, we have a couple different ones here. As we scroll down, we can pin these to our section in the user data and add them. From down, any recent conversations that your customer had. So say, for instance, your customer reaches out pretty frequently, all of their conversation history will be this appear in the recent conversations. Any user notes, if you wanna add notes about the user, say, for instance, they reach out frequently, and you can add that to the notes. You can tag your users. You can also tag conversations here on user tag section. If you wanna group your users for a specific segment, you can create a tag and then tag the user specifically to have that segment. Use your segments, recent page views for tracking the customer if they're going two different pages on your website, similar conversations, and company details. You can also click and edit this, and you can pin and unpin sections if you'd like. Now one other thing I wanna show you, and this is really useful for reporting and if managers wanna dive deeper into the conversations that they're having. We can go into the report section. We can overview all our reports for conversations, so how many conversations you're having, anything that Fin is involved with, chatbot on the teammate replied, pending, anything that Fin's resolved, and your teammate replies. So you're getting a ton of great data here. Now going back to the inbox, one thing that I do wanna mention as well is FinCopilot. Now FinCopilot is an internal version of Fin where it's not directly communicating with your customer. It's actually communicating with your team. So for instance, if I want to highlight a conversation topic and ask, how can I ensure all my teammates use two factor authentication? We can ask Copilot, and FinCupilot will go through the data that you have ingested or ingested, and come up with an answer to ensure how your teammates can use two factor authentication. Now you can preview this before you go and send it to your customer. Once you read this and if you're ready to send it, you can add it directly to the composer and then send it directly to your customer. Right? Another really cool thing I wanna show you that links into our inbox is our SaaS dashboard. Exemplary is just a mock software that we use to kinda mimic your website. So when you come up into your website and you install Intercom Messenger, there's a little widget in the bottom right hand corner you should click on, and this is how your customer sees the Messenger when they come onto your site. And you can add and customize these as well. So if you'd like to ask a question, if you'd like to add help center data, or create a support ticket, Your customer can go and ask a question from the messages tab or search your help center without having to leave your website. For this, I wanna show you how, a customer can create a ticket. So if you add the ticketing option to your messenger, you can click on create a ticket. And get email notifications. You can add Nope. And then your customer can send a ticket directly into your inbox from your messenger. The ticket data will be hosted in the messenger if your customer logs in. And when they come back, they'll be able to see all of the ticket data. But once they send that through, that conversation will come into your workspace. And if you have any routing set up, you can write it to a specific place, but for now, it's going into our all folder. So once your customer does that, you can reply back to the customer and work directly on the ticket from there. Oh, fantastic. That is concluding our demo for the intercom help desk, and I'm gonna pass back to Elena to tell us more Perfect. Thank you so much, Kendall. So Fin. I'm gonna share my screen one more time. And, yeah, now we are going to meet Finn. Yes. So, Finn is our next generation AI agent that delivers human quality customer service and is now handling some of the most complex customer queries out there. Today, it's not enough to simply deploy your AI agent to the masses anymore and hope for the best as your customers need to continue to involve. Yes. Same is applied, yes, to your customer, to your AI agents, basically. This is where the AI human hybrid approach comes into play. The AI agent relies on constant adjustments from you and your team to reach its full potential. The beautiful thing about Fin is that it can work for you twenty four seven. It doesn't get sick, it doesn't go on holidays like your team. And with Thin, yes, the perfect is is possible all the time. So nowadays, most AI agents can handle the very easy stuff, like simple FAQs or basic how tos, and that is the kind of minimum, yes, that you can expect from, AI agents. But, what is the real difference between an AI agent that resolves, let's say, 30% of queries and another one that resolves, 60% or more? It is how well, it handles the tough high value queries. Things like, refunds, escalations, and those, weird edge cases that usually slow everything down for you and your team. And that is, where Fint stands out. It is designed to go beyond the basics and take on the hard stuff, the things that actually eat up most of your team's time. Fint can pull in the right data from different third parties. And if it's necessary, it can take the right steps and complete, the whole workflow from the beginning to the very end, without any human involvement. And now I will share a bit on how Fin, has been, evolving. So, Fin was actually the first AI agent to build specifically for customer service. We launched it, roughly, three months after ChargeGPT came out. And here is a, like, a key thing. Yes. It resolves more than any other AI agent out there at the moment. And right now, it performs at 67%, resolution rate and, delivers their highest quality answers. This is not just, down to the underlying AI model. We are constantly adding some new features that makes FIND more capable and more useful for our customers. As you can see on this graph, we've managed to improve Fins resolution rate by about 1% every single month, for the past sixteen months in a row. We are not slowing down. We fully expect that, to keep climbing in the future. And now I will hand it over back to, Kendall. Kendall will show you how Fin AI actually works in practice, And I'm gonna share the last poll, about what is your experience with AI in customer support at the moment. Okay. I said thank you again. Hello. Alright. So we're gonna head back over to Exemplary, our mock dashboard, and show Fin in action. Now, we're gonna start with an informational query. So an informational query is something that is running static data that can be completely answered numerous times over and over with help center documentation. So for instance, let's say we want to ask them, I'd love to know more about your plans. Can you share more details? You can ingest data about your plans into a health center article or link it from your billing page and go directly into Fend, and Fend will be able to take this information that the customer's asking and answer it accurately. So at the same, Fend is coming and telling our customer about the different plans that we have, the starter, the pro, and the enterprise plan. And you say if that's what you're looking for, customer can say yes, and this will be considered a resolution. A resolution is when Fin is able to completely answer the customer's question. You don't have to route it to a teammate, and then you'll be charged on an answer because Finn is answering the question. Now Finn can also handle other things like personalized queries, and personalized queries are things that require a bit more in-depth detail about the customer's data specifically. So I can ask Finn, what subscription do I currently have? Finn can go and look into the back end on the data that you have ingested or connected via data connectors. Look up what subscription I currently have, and then Finn can let me know. Currently, I'm on the pro plan. This includes 25 users and five additional users total, and it costs about $800, which is fantastic. Now Finn can do quite or a couple other really cool things as well because it works directly in the inbox in Messenger, but it also can go over email. Now it takes about two minutes for a message to be sent and delivered via phone because Internet providers give you a two minute window to pull back messages. So I prewrote this message asking the same question, what subscription do I currently have? And then from our exemplary SaaS platform can also reply over email telling us what subscription we currently have in our plan, which is great. Now if we wanna go back and go into our Messenger and say, for instance, you have customers who are reaching out and may speak a different language, they can also reply in different languages as well. So for instance and my Spanish is terrible, so I'm not gonna be able to pronounce this, but it roughly translates to, I believe excuse me. How can I make sure that all my teammates use two factor authentication, which we asked Copilot previously, but we asked Spanish this time? Now Finn can take the information that has. You can see that it's sent in Spanish because we sent it in Spanish via exemplary, but it shows in English because my workspace settings are set to English. My device settings are also set to English. Now reply back. I sent in Spanish. And as we reply back in the inbox, it comes back in Spanish to the customer. It's been working in different languages. It works over different channels. And now I wanna go into the Fin flywheel, which is how you set up Fin in your workspace kinda bring all this magic together. So we're gonna go to FIND AI agent and go over to FIND service. Because as I said, we have sales and service, but we're working directly on service. Now FIND service works in a couple different forms. So it's train, test, deploy, and analyze. In our train section, we're adding our content. So if you wanna ingest any content like public articles, internal art internal articles, snippets, or sync our website, this is where all of this goes into FIND for it to learn and be able to respond to your customer. You can set specific guidance. So say, for instance, you wanna change Fin's tone of voice to be an enthusiastic product champion and to not use emojis. You can change the communication style, add context and clarification. You have your content sources here, any spam guidelines, or any other guidelines you want Fin to follow to make it more in sync with how you and your company reaches out and speaks to your customers. Now you have attributes for any attributes that you wanna add, escalation process. So say, for instance, you want to route greet customers to a specific team, you can have Fin do that internally. And then any procedures that you wanna take, procedures, it kinda allows Fin to, like, do more complex things, like is that refunds or go in and look at specific data. Excuse me. You can troubleshoot billing. A ton of different things you can do with procedures. It's a little bit more technical, and you'll be setting up data connectors. This is when you go in and connect your API system into FEN. You're gonna be able to look through that data specifically about your customer or your software. Now when we look to the test section, this is where you test and troubleshoot FEN to make sure that it's giving complete accurate answers all the time. You can see an answer status and an answer rating. Excuse me. And as you're going through and you're testing Fin, you can kinda make adjustments to see how Fin is kinda operating and where you can kinda, like, update your data or update any articles that you have to make it more accurate so Fin can give you more accurate answers. We're going to our deploy section here. You can deploy Fin over chat, email, and phone. So say, for instance, we wanna set a simple automation, we can set Fin up to work directly when the customer comes in, or you can do an advanced deploy via workflows if you wanna get more tailored to how you want Fin to work. Then we have the same options for email and phone. Then in our analyze section, you can analyze Fin's performance to see where you wanna make any changes. So you can see the automation rate, the involvement rate, and the resolution rate that Fin is handling, then your CX score positive CX score reasons and negative CX score reasons. You can see your performance funnel to see how Fin is actually operating and if you need to make any changes there. You need recommendations that Fin can make. So recommendations can be Fin adding content to kinda help you tailor to what your customers are asking. You can edit articles for any content gaps, also to see where you guys are missing data, so Fin's covering your blind spots. You have the topic explorer to see the topics that Fin instead of covering. So for us, subscription plans is the biggest one. And you need advanced analytics, plan details, and then two factor authentication. You can monitor any trends that Fin is seeing. So say, for instance, your customers are asking about pricing or onboarding consistently, Fin can monitor trends, and then you can monitor all of this from the monitor section. And that is Finn in a nutshell. So I'm gonna pass back to Linda to bring us home. Thank you so much. Perfect. So yeah. Thanks a lot, Kendall. Now, yeah, when it comes to making this shift to AI in your customer service, you really need two things, basically. The best product and the best partner to guide you through it. As Kendall said, Affin is number one agent out there. And because it delivers the highest quality answers and handles the most complex queries, not just easy stuff, it is built for this new, era where AI isn't just helping your team. It is actually transforming, yeah, the way they work. And, with Intercom, you're not just getting great tech. Yeah. You're getting the partner, with over ten years of experience, yes, in customer service and deep, deep AI expertise. That's why we are the most trusted name in, AI customer service transformation, at the moment. And, let's quickly look at some of our customer stories and what our customers are saying about us. We didn't have to cherry pick the results on this slide, so to only present you with the best performance statistics. But, literally thousands of customers are using Fin at the moment, and they are seeing dramatic, dramatic improvements in their customer service operations. Their customers are happier and they are getting immediate accurate answers over 60% of the time. And overall, CS operations are happier as well, doing, far more with the same investment and reducing costs in, many areas. Critically, yes, human agents are happier right now with FIND and Intercom because they can work on higher impact tasks. Our customers, they have moved their human agents away from repetitive questions and onto new channels doing customer success and spending more time on tasks that really require some human involvement. Here is a closer look at, one example, and that is Entropic, our customer. So they are, a leader in AI world, which says a lot, I believe. They choose to implement Fin, instead of building their own AI agent and saved roughly 1,700 support hours in just one month. So they've achieved over 90% fee involvement rate in every conversation and roughly 50% resolution rate in the first month. Right now they are at 62% resolution rate with us. For a fast growing complex tech company like Entropic, that's a massive boost, yes, in scalability and quality of support. You can find the full customer story of Entropic and different other customers as well on our website. And I would really recommend you to have a look at it because there is a video interview with their head of product support there, in which, which you might find very useful maybe for for your support setup as well. Perfect. And, that's it from our side. Yes. That's the core of what Intercom and FAIM can do. We'd love to open it up right now for, questions from your side in the q and a section. Let's try to keep it, relatable, yes, to what we've just shown you. Any, like, very complex tech questions, we are more than happy to take it offline. So, do we have any questions? Ron, sometimes we have issues with software which affects many customers who are using the certain integrations. Confin recognize that multiple tickets are about the same issues, so we can rely all at one we can reply all at once when the issue is resolved? We can do it with a tracker ticket. Definitely. Right? Perfect. So tracker tickets are kinda when you have specifically for that instance. So say for instance, you have a bug that's affecting multiple customers, you can create a tracker ticket and group all of those tickets together in the tracker ticket and respond to the tracker tickets so all of your customers are getting responses for that specific issue at once. Okay. I would like to clarify whether thin can be set up to use different content depending on where the user is. For example, on public sales pages, it would use marketing service and sales related information. While inside the platform, it would use separate help center onboarding support content. Mhmm. Okay. It's a very long question. Yes. You can do that. You can set up different audiences. Yes. And depending on the audience, Finn can provide different kind of information from the knowledge center. Yeah. We can help you to set it up. If you have a trial now, yes, just write us a message, and we will help you to set up different audience. The second part of the question, I would like to understand whether FIM can always be the first point of content where it tries to answer the user first. And if it can't help properly, it then pass the conversation. Yes. This is how thing works. Yes. It it can be like the first line of support for you. Okay. The section one can also show an email conversation being translated. Does an agent have to approve the response before it is sent out? In the section where we're showing the email. So for Fin AI agent, an agent doesn't have to approve it. Fin is trained on information that you give it, so Fin will be able to answer directly from when the customer reaches out. In Copilot, then would take that information and answer the question, but then the agent would have to kind of read it and approve it. So for AI agent, it doesn't have to be approved. For Copilot, you can actually read it and approve it before sending it. Any other questions? So for Copilot, it doesn't alert you. You ask Copilot the question, and then Copilot goes in and gives you a response. So it doesn't automatically send in Copilot. Copilot will give you the response, and then you can copy and paste that and send it to the customer. So it doesn't necessarily alert you. It is, you know, written in the way where you have to, like, actually copy and paste it. Copilot, it already works with conversations that were reassigned to the human agent. Yes. So when the human agent, got a conversation, they use Copilot, yes, to to generate the answer. Can you rack a website? I don't necessarily know how to answer this. Can you provide a bit more context about what you're looking to do to Ralph Lauren? Or is it? Is it easy to connect to your own API? Yeah. Yeah. Like, relatively easy. We have API documentation on the website. Yes. And yeah. For For data connectors. Yes. So So with Fin, you are ingesting your website content into Fin. So you can add, like, a URL from your website as, like, a content source for Fin. It'll scrape your website for any information that your website has and then then can use that information to answer questions. So, yes, technically. You can you can use different sources for FIND. Yes. You can you can have, like, a content from your website. You can connect, other software, for example, Notion or anything else where you store your knowledge center or articles in Intercomium. But website is totally possible. Any other questions, guys? Anything else in your mind? Okay. Give it a couple of minutes in case someone's furious to tap in. So if we have high volume and resolved conversations, the price we fit the price for the same resolutions are always gonna be 99 something. It depends. Yes. So let's say if you're on self-service, yeah, which means you pay monthly, the price is fixed, but if you sign an annual contract with us, yes, we can potentially discuss that, but it is all based on just, like, individual negotiation, basically. So we need to talk to sales about that. Yes. But if you're on a monthly payment terms, it's always 99. Any other questions, guys? Alright. If there are no okay. There are more questions. What about the leg legally side of using the customer information which is available in API? How do you mean like, are you asking about the legality of using customer information to pass over via API? Or a bit more context here probably if you needed it. For that, I would probably recommend reaching out to our team via the Messenger because that goes into a little bit more technical side. So our engineering team will probably be a better source for that about picking and choosing which data points you can pass over and not pass over. Yeah. Ron, and regarding I I cannot reply to you for some reason, so I just I just, pasted here. We have a trust center, so you can, maybe check about the compliance there. Yes. Perfect. Yeah. If there are no other questions, yes, what's going to happen? Yes. First of all, you will get a recording of this webinar. Yeah. You can share it with your colleagues if necessary. And, yeah, if you want to start the trial of Intercom and you're already on a trial, feel free to talk to us if you run into any technical difficulties with the setup or any questions about prices, anything in your mind. Please write in the chat, talk to product specialist. We will pick up the conversation with you and help you evaluate Intercom individually. Thank you. Thank you. Bye bye.