Name: Manage Peak Season Volume with AI Agents
Uploaded: 2025-09-24T17:18:10.469Z
Duration: 1 h 48 s
Description: Manage Peak Season Volume with AI Agents

Transcript for "Manage Peak Season Volume with AI Agents": Hello, everybody. Hello. We got everybody up on stage. Welcome to our webinar on how AI is transforming ecommerce and how brands are using it to gain an edge during peak season. My name is Mark Iafreight. I'll be your host and moderator for today, and we've got a fantastic speaker lineup for you. We've got Tim Kushner from the The Nest by Concentrix, Chip Lay from Fullstory, and Intercom zone, Ruth O'Brien. And I'm gonna let themselves, introduce themselves in a second. But before we dive in, I just have a few very quick housekeeping items. Say hello in the chat on the right. Let us know where you're joining from. I see Tracy's already done that. There's a q and a tab there in that same panel. Go ahead and submit questions that you have now or at any point. You can also upload others in there. We'll have time at the end to answer those live. I've listed a bunch of really useful resources in that doc section, which you should check out when you have some time. Obviously, we're recording today's session, and we're gonna be sending a recap, some of the key takeaways, and a link to the recording to everybody here tomorrow morning. And you'll also notice there's a poll tab. We have a poll open right now, so go ahead and complete that. I'll just toss it up on stage really quickly. It's what's the most pressing challenge for your peak season this year. So go ahead and select one of those, and we'll have some time to kind of go through that as we go through today's webinar. So all the housekeeping out of the way, love to meet our speakers. So, Tim, why don't you kick us off with a quick intro? Thank you very much. Hi, everyone. My name is Tim. I'm based in Berlin, Germany, and I'm so glad to be here. I'm part of The Nest by Concentrix. We are an intelligent transformation partner and the leading agentic AI company worldwide. We have companies across all sectors, design, build, and run customer experiences, making the right AI decisions, get AI ready, and scale AI deployments globally, either in by either by partnering with great companies like my fellow speakers here or by deploying our own product suite. Great. Thank you. So great to have you. Thank you for taking the time. Chip, do you wanna go next? Yeah. Great to, meet everybody. Excited to be here. My name is Chipotle. I'm on the product management team at at Fullstory. Yeah. We've been on a journey. You might know us as, our roots in session replay, but we are now really a behavioral data platform that's powering, a lot of what's happening in the AI world when it comes to digital body language. So excited to be here to get to chat. And, yeah, looking forward to the conversation. Amazing. Thanks, Chip. And, Ruth, you wanna take us home? Hello, everybody. Ruth here coming at you from Dublin, Ireland. I lead our AI support team here at Intercom. So what does that mean? That's the team who implement Fin and all things AI for our own customer support team. For I'm actually at Intercom for about ten years and for most of that time I led the EMEA frontline support team and then I moved into this more focused role now all about AI implementation for support. So like any other support leader I care about things like cost and efficiency and customer experience across the board but I have the added fun and pressure of the fact that my team implement Finn to show off to our customers who we want to buy Finn and if they experience a poor interaction with us with our Finn implementation they're not gonna wanna buy it themselves right? So it's like that add a little bit of pressure and fun that comes along with my role, but I do absolutely love it. And I'm really excited to chat with you all today about all the really cool things that Finn can do in the ecommerce space. Love it. Thank you. Yeah. Like you said, no no pressure on that. Well, thank you everybody for being here. Really excited. We're gonna go kind of right into some of the content. Recently, I think a lot of people probably saw the Shopify CEO, Toby Luecke, talked about how AI is now the baseline for running and growing, kinda like retail and ecommerce businesses. And what that really means is AI is becoming the operating system of retail and peak season is where a lot of that gets tested. So today, we're gonna be kind of focusing on three core areas. One is AI for the kind of traditional post purchase support, something that's a bit more proven. We're gonna spend a good amount of time talking about AI for prepurchase conversion, so kind of the growth frontier of AI in this space. And then we're gonna spend a little bit of time at the end talking about how you can actually get started quickly, in time for peak season. So, Ruth, I got this first question here for you. I know you've worked in retail, in the support side before the AI era. So I'd love to hear from you. What feels most different now, and what do you wish maybe you would had back then? Yeah. I used to work, in customer support for Apple and you can imagine, like, launch time, for a company like that. I also worked in gaming for Blizzard Entertainment at one point and if we launched, like, new functionality or new games, the influx of new customer interactions at the time it would be like in a call center was wild. So everybody would be battening down the hatches, there'd be loads of new part time staff being added in, people will be working overtime. It cost the company a fortune in scaling and also like paying folks for overtime as well. And no matter what, you know, how many people you'd add to it for a really popular launch, you would just be like dealing with backlog, you know, on fire, customers waiting for, for responses for a very very long time. And now with AI agents that can look completely different. We just had a launch at Intercom. I know we're not ecommerce ourselves but we had a launch at Intercom a few months ago and because we had ourselves in order with all the content that we needed for our AI agent, we were able to see a 70% resolution rate on that new product which is our insights product. And if you imagine that in the thousands right, say we get a thousand conversations in one day and Fin is resolving 700 of those, that's 700 conversations that's not going to our human team. So if I think back to my time when we used to do game launches or product launches in companies I used to work in before, It would have been a complete game changer to be able to have something like Fin answer so many of those common questions because let's be real like most of the questions that come in around a launch are similar things you know. Questions about it depends on what it is. Right? But questions about subscriptions, questions about the product. So much of that now can be handled by an AI agent and you don't need to scale in that same way whether it's temporary temporary scaling or long term scaling of your human support team. So it's a different space now. We had another launch recently where we were overhauling our reporting at Intercom and it was a really big deal for our customers. It was quite a complex launch but we actually on the human team barely even noticed it. We enabled our team. We gave them all the training that they needed but they barely got any conversations about it at all because we had Fin so empowered to answer these questions. So we were doing nothing. We were like, woah. We meant maybe we need to ask people to work late or whatever it is, and we didn't need to do it. Like, so for the human support team to be able to barely feel a pretty mega launch, like, that is game changing stuff and that was not the case back when, I was doing launches earlier in my career. Yeah. No. And that's a it's a fantastic segue because I I had a question for Tim. Kind of in that same vein with the rise of AI, how have you seen the needs change, for support teams, and and how have you seen that kind of evolve? And, how are leaders kinda preparing for peak season now differently than compared to past years when it comes to, the teams themselves and maybe the human component there? Yeah. It's, fundamentally different, and I can just, remember what Ruth was saying. I mean, in the old model, it was really about people, people, people. Right? The scaling big teams absorb volume. But now, I mean, deploying FIN or our IXLO suite, we we look at automation resolution rate of 80%. Right? So leaders can cover a huge share of the routine work with AI. That means peak season are less about throwing people at the problem and more about having smaller expert level roles roles ready for complex, very nuanced cases that the AI basically escalates. And this also has an impact on the off season. Right? Because leaders should use their time to tune their tech stack, their workflow. So when peak hits, teams can basically just stay much more chilled, and this has an effect on not only the team itself, but the overall company's shareholders are happy, VCs are happy. Everyone is gonna be happy. Yeah. Absolutely. And, the the that comes with, like, bums and seats. It's the the kind of a specialization now for these certain types of roles. We've talked a lot about that as well, these kind of new new types of areas where support teams are working. Again, great segue. We wanna kind of, like, talk a little bit about what we call the the proven use case for for, AI and support for for ecommerce. And, this is kind of where the kind of rubber starts to meet the road. So, Ruth, another kind of question back to you. You touched on this a little bit, but where do you kind of see most companies beginning, where they implementing AI agents for customer service? And for them, what is that, like, getting started phase typically look like? What I would say to anybody getting started, and I see it with our customers all the time, is, like, get into that low hanging fruit as quickly as possible. Your team and probably, like, if you're a support leader yourself, you probably know some of the most common queries that your customers are asking you about and you have an idea about what your knowledge base or help center is like. Like, does it actually answer those questions? And they're the ones that you can really get started on with Fin very, very quickly. You just need to make sure that you have the content that Fin ingests up to date for your most common questions and then off it goes and it starts answering those incredibly well. And so that's definitely like a starting point. And then pretty quickly you can start scaling on that. You just need to make sure that you're putting in the time and attention and dedication to, again, like, the content. Keep bulking out the content. Empower and enable your team to keep, suggesting new content as they get new questions from customers, and you'll just see that resolution rate start to absolutely skyrocket. But another really cool thing now especially in the ecommerce space is that the technology is improving so much that quite fast after you nail a lot of those informational queries, you can start doing things that are a lot more complex that humans would have always needed to have been in the mix for before. And I know that we're gonna talk about that a little bit more as we move through the discussion. Yeah. Absolutely. And I actually that's a, good kind of lead in. Chip, I actually wanna throw this one to you. Like, Bruce talking about kind of this foundation level and, you know, we know, you know, good data in, good data out. Like, there's an important level of foundation when it comes to content for an AI agent, but I I would have to assume a lot of that is kind of similar for, the kind of, like, the behavioral data lens. So when it comes to kind of getting you getting ready to to launch an AI agent for support, do you have any kind of, like, thoughts around best practice for getting started more from that kind of data, behavioral data lens? Yeah. Absolutely. I mean, data quality definitely matters immensely here, and having a complete view of your customer experience is critical. You know, really understanding how your customers experience your site, where their journeys take them, where the rough edges are, where people are getting stuck in loops. Having that type of clarity and that kind of empathy of understanding that journey allows you to really focus in on where to optimize content. And it it really is a nice interplay between understanding the conversations that are happening and then the analytics of what those customer experiences are and connecting those two together can create, your, you know, a strategy and roadmap on how to continue to scale and optimize effectively. Yeah. Absolutely. And and, Tim, to to you, when when y'all are advising companies that are maybe just starting their AI journeys, how do you go about doing that kind of what what is your starting point there? I mean, I would love, like, who said it, to always go for the low hanging fruit because that's most fun. And you see immediate results, but Chip is also totally right. I mean, I usually give an answer that my clients don't like, and I ask them, okay. Are you actually ready for AI? Because what we are seeing, and we serve over 2,000 clients globally, AI maturity is so, so different. Most company thinks they are far ahead are basically getting started right now, and the common blocker is outdated systems, disconnected systems. Companies, usually, when they grow fast, they develop silos. So CRM warehouses, we found tools that don't talk to each other. And if you put AI just on top of that without having the clarity that Chip was talking about, you create chaos. So So this is why we think, and not saying everyone has to do that, but we advise it. And agentic readiness assessment is basically critical. Map the customer flow. Align the stakeholders. Check if your data and systems are integrated because most failed pilots basically happen because leaders getting excited and they start to overpromise because, before those foundations are in place. Yeah. And and I would have to also imagine part of that is sometimes the the the leaders in place, they they may not even know the the kind of status of a lot of these systems because they may not be in the weeds yet as as other folks. Yeah. Everything seems very clear. And, I mean, that's probably the coolest thing about this development, but, unfortunately, companies are not built easy most times. Yeah. Absolutely. And then there's the the the tech debt after you've been growing and things can get can get messy. But we've there's been a lot of discussion around failed pilots, and and I think a lot of the ones that I've read kind of talk about these very large, typically what people would call bloated organizations, but I think the same thing is true potentially for some even smaller or midsized companies. I don't think people sometimes realize that, there's a certain level of cleanliness that you need. I know Ruth talked about it from this kind of, like, knowledge management lens. So making sure that your external documentation is is up to snuff. It's got, comprehensive coverage. It's accurate, and you have the right teams in place. But that's also the same thing from the data the data analytics lens, the the behavioral data lens Chip talked about, and even just generally making sure that you have the right integrations in place. And maybe just just one thing to add, Mark, and sorry, I'm too for blowing up your agenda here. But, I mean, a lot of companies like to deploy Shiny AI features directly to their customers. Why? I mean, get yourself AI ready first. Deploy it internally. We see the most really relevant improvements in equipping workforce with AI. Enable your people, enable really your frontline advisers with AI First to, reduce, search times to be able to have more time to, handle queries with empathy. And this is really the one thing that I would focus on before basically rolling it out to your customer base. Yeah. And that's actually a great quick Ruth, I know I know that we have, that they were the kind of, Copilot. It's kind of like a way that people can be using FINS internally for their teams. And I know not only do we use that, we've seen other companies use that. But do you have any kind of, like, thoughts on on kind of that kind of more AI being used as a human Copilot perspective? And, do you see sometimes customers or businesses, starting with that as, like, an initial pilot rollout before they turn it on live for, for their customers? Yeah. We did that before we launched FINS. So pre FINS, Intercom built out some inbox features. They're called, like, AI assist features in the inbox, and that was even before our Copilot functionality. But that was, like, in early days because, like, we moved really quickly after ChatGPT was initially launched back in 2022 and started implementing functionality in the inbox that would allow people to do, like, some of those classic features which now we're like it feels like ages ago. It's like summarize, expand, make it sound more professional. But having all that before we had even launched Fin as a product was so helpful for our team to get their heads wrapped around how this type of technology actually works and they could start to trust it a bit more and that helped us then when we were launching Fin and the customer facing end of things that we had a bunch of folks on the team who actually understand how the basics at least of AI works. And that means that the people on our team are able to to understand like if Fin if something goes wrong with Fin, they understand why, they understand what happened. Was it content? Was it knowledge? Was it whatever workflows were set up? Was it how the customer asked the question? Was it how we wrote our content? You know? There's, just so much to gain from making sure that the folks on your team actually use and understand AI so that they can help when you deploy AI out customer facing. They can help you make that better. Yeah. It's a way to, number one, just build trust with your human team around around, like, the the capabilities of AI. It's also a great way to get them to understand the the basic functionality. How do I actually use this to to derive a lot of really solid value? And then I there's also that that kind of tactical use case of knowing, hey. We don't have good coverage in certain areas. Right? We there are certain things that we're maybe missing that we need to go build out before we start to to deploy this live to our customers. Yeah. If I can Oh, go ahead. Sorry. Go ahead, Shiv. I was gonna just we're hearing the same thing from, the Fullstory side where a lot of our enterprise customers are asking the same questions, which is where do we start with our workforce, and where where are the workflows that matter most? And, you know, having that, that's actually led us to design and and deploy a product we call workforce, which is Fullstory for the, employee, for the enterprise. And, having that data and having that ground truth about how people work is so important to then understand how to build, and deploy AI agents, whether that would be Fint or other agents like technology, to understand where you're gonna have the biggest impact and then measure and improve. So I'd highly recommend the same thing, and it's it's exactly what we're hearing as well. Fantastic notes. I love love that. And, just a reminder for folks, we have that poll open and that poll tab, you see a little red dot there. Go ahead and, fill out that first poll. We'll get to that after the the next section where we view some of those results. But for now, I wanna kind of transition us into, kind of a the growth frontier of of AI. Talking about, like, kind of AI for this prepurchase conversion for sales kind of use case, bit more of an slightly more emerging frontier. And so, like, Ruth, you've touched on this a bit already, but once teams start to see success with AI for a lot of their frontline support, they quickly realize it can do a lot of other things. Right? It can drive revenue, not just improve the the general customer experience. So, I know we also recently published a case study on one of our new customers, WHOOP. They saw a lot of really strong results using AI to kind of drive their inside sales team. So can you maybe just give us a a quick primer on that and then maybe broaden it out a bit to let us know what you're seeing at Intercom and maybe with some other customers? Yeah. What we did is so cool. So they had a three person support team, which is quite small. Right? And this is exactly what we want from folks deploying AI agents is that you stay small and you don't have to have this linear scale that you would have done before with your massive spikes in business. So during one of their launches they saw a like 20 x spike in their chats which is pretty intense for a three person support team. Right? But Finn handles about 68% of those questions and it actually this is the incredible part. It actually doubled the sales attribution from the inbound chat that they had. So they found that if they as humans were getting back to people quickly, folks were more likely to convert and actually purchase rather than sitting around waiting for their question to be answered for ages. You know, they get distracted customers and go off and potentially go to some competitor and who is answering them faster. Right? But, because Finn was actually now answering in real time so many of these questions, the conversion for the sales funnel then was, doubled, which is absolutely incredible. So it's really cool to see Fin going beyond this classic, you know, support use case, I suppose, and actually affecting the bottom line and the revenue of the company overall. And at Intercom, a way that we're doing this on my own team is we have Fin selling Fin. So if a customer asks our Finn a question and at the end of that, Finn resolves it and they say, yeah. That helped. Finn reaches back out to the again proactively and if they don't actually have Finn so it's very much if they don't have Finn purchased on their account, Fin will say, like, I see you like what you saw, that you got a good interaction with me, and I answered your question. Would you like me to help you get set up, for your own business with Fin? And then it will go through a flow where we get them set up with a demo for the team. But the whole point about this is using AI to sell more AI. So it keeps coming back to that space of, Fin going beyond a classic support use case and then support teams, not being just like a cost center to the business like they have been in the past and actually becoming a value add for the business overall, driving revenue, driving product adoption. It's really cool to see these things, on the move at the moment. Yeah. It's it's it's that shift from this reactive world, engaging, like, in the moment when somebody has kind of triggered something to a more proactive approach, which is really exciting. And, have there been any other kind of, like, customers maybe kind of in the the, retail, ecommerce space that you've kind of seen, doing some creative things? Yes. One of our customers, Nuuly, who are a clothing rental brand, they had a high volume topic, which was like pausing subscriptions. So they have, like, I think it's a monthly subscription process and customers can reach out to their support team, ask to pause their subscription. And humans always have to do that before, right, because it involves going out to a different system and the humans making a decision about what was gonna happen, pressing the buttons and doing whatever they needed to do and then going back to the customers to tell them that they had done it or ask a bit more qualifying kind of questions, in the meantime as well. But, now they're using Fin to own that process end to end. So this would have been kind of more of like a a kind of billing or, like, type use case, you know. Again, it's not like classic classic customer support. It involves more, like, financial, sometimes more emotional conversations. And customers like newly now are letting Finn handle that from the very start all the way to taking the action and coming back and telling the customer it's done. And it never even has to touch off a human at all. So it's really, really cool in the world that we're in. Yeah. It's it's a really exciting trend. The increased complexity of what AI can handle and, you know, AI moving upstream into this, kind of calling, like, a proactive prepurchase engagement that directly drives conversion. So, Chip, I wanna kinda toss it to you. Can you first maybe just give people a sense of what we mean when we talk about the proactive prepurchase engagement when it comes to AI agents? Yeah. Absolutely. We're we're having this discussion with all of our major retail customers, which is how do we move up funnel and what is and so what does that mean? Well, a lot of, a lot of, you know, AI chat, interactions agents are can be post purchase. It would be like, you know, where's my order returns, things like that. For, for our customers, you know, having the right data and context for that is really coming from other back end systems. But, you know, when it comes to further up the funnel, I have an analogy I think is kind of, like, a helpful way to frame this. So imagine you walk into a brick and mortar retail store. You know, you're not gonna have somebody from, you know, store associate run up to you right when you walk in the door and just get in your face, be like, hey. How can I help you today? They're gonna watch you. Right? They're gonna, like, see you come in the store. They're gonna watch you move through the aisles, kinda check prices, maybe your comparison shopping a couple of different brands, and then they're gonna come in at an appropriate time and ask the right relevant question based on your body language. Well, that's the same thing here. Right? How do you recreate that digitally? How do you find the digital body language that helps, retailers understand in the moment intent? And what you know, another big problem, that I think in the personalization space in general is that more and more, you know, you don't know your customer when they're on-site. Like, the vast majority of retail customers are anonymous when they're there digitally. So how do you understand their in the moment intent and then translate that into proactive engagement? To Ruth's point, like, doubled the, attribution of sales because it was the right message at the right time and it was appropriate. And so using this kind of, digital body language, understanding when somebody's maybe comparison shopping, when somebody's hesitating or maybe having a hard time making a decision, or when they're frustrated, when they're trying to add something to cart and it's out of stock. Maybe they're getting error message. Triggering, then actually engaging at the right time with the right message, critical. Right? And then really creating an engaging user experience. Yeah. I think everybody's had that experience where maybe you're you're online and you're going back and forth between a couple different products or you're constantly clicking on that, find my size, button to try to figure out, like, what what should I be getting, being able to kind of take that take that language and then being you know, same thing as somebody holding up, like, a shirt on them when they're, like, trying something on to see if it'll work. Kind of being able to proactively nudge people and help them in that moment when the the intent is high. That's right. 100%. I mean, it's such a better, more human way, I would think, than the kind of traditional approach of, you know, right when you hit the site, you know, 15% off. Just give us your email address right now. It's like, give me a second. Right? Yeah. So yeah. How do how do you create that? You know, it's just right at the right time, not too early, not too late. Like, you wanna engage, in a timely way that feels very, very appropriate. I think the, you know, the other thing that, we can learn a lot from these conversations that happen, in the intersection of those digital journeys. So, you know, kind of combining analytics, really understanding the customer journey. Every single conversation is, is a is a signal that then can help you then improve your customer experience. And so getting that feedback loop and that flywheel going so that, you can really kinda decrease more of the friction based chats, and you can help you know, maybe you're moving the site around because you're getting a lot of, you know, confusion about a certain feature or certain flow. That's gonna allow you to, you know, point that energy towards, you know, more proactive tactics where you can really drive conversion rate investment size. Yeah. Yeah. And that the those those signals that you're able to get, the being able to identify that friction, you can then almost kind of categorize some of those. Right? There's if it's if it's somebody asking a question afterwards, maybe there's a gap in your content. If it's pre purchase and somebody's struggling, maybe you have to figure out, maybe we have to redesign some of our UI or maybe we have to make our sizing a bit more clear. Right? So there's there's a lot of, like, viable insights that can not only just be applied to the kind of, like, customer service or presales use case, but even things like the general experience of a website or how you've designed something or the information that you're showing to people. Yeah. Yeah. 100%. Absolutely. And and, Tim, do you have any kind of, like, thoughts on this or anything that you've kind of learned similar from, some of the customers that y'all have have worked with, at at The Nest by Concentrix? I don't know. It really I mean, I totally agree with everything that's set. And, I mean, what we are seeing is that context has came right. So, I mean, you can definitely automate most of the repetitive work nowadays, and I think the group example is a brilliant example for small scale teams going into a peak season. But, yeah, I think, the nudging piece that Chip just mentioned is something that most companies have not on their radar yet. So when are we triggering the right signals to either go a different route or actually, support our people to upsell, retain, or escalate in a proper manner? Yep. Absolutely. And, I was gonna really quickly review the results of that first poll. If people wanna keep dropping in, notes, they can. But, we were just talking about what's the most pressing challenges here, and it looks like, actually kind of spanned out. It was surprising. A lot of folks talking about managing spikes and support volume, SLAs. I know that that's super important. Proactively drive sales was a big one, which is really interesting to see. And then, obviously, we got that kind of loyalty CSAT angle as well there. But, based on everything that we said, I think it's regardless of what your biggest challenge is this year, there's going to be some level of of, support you'll be getting from an AI agent to kind of help tackle some of these problems just depending on where you wanna get started or where you wanna kind of apply, some of your your resources. That being said, I'll just quickly open up the second poll. People can feel free to drop that one and, responses in there as well. And, as we get a little bit towards the end of the content, we'll review that before we go into the q and a. So I I wanna kind of switch gears again a little bit and talk about this AI plus human dynamic because we've talked a lot about that, but I think it's, really important to to touch on. So, Tim, I've got this one here for you. With with AI handling so much of the repetitive volume of support queries and even outside of the more complex tasks or even more outside of the proactive side, just that pure repetitive volume, how are teams using that extra time they've gotten back, and how have the roles on the human side evolved? I mean, with AI taking care so much of those repetition support teams can basically focus on higher value work. Right? And conversion rates wise, a, like Chip said, when nudging is in place, but also when human bandwidth is freed for what really matters. So escalate escalation, especially in peak seasons, that are not handled by AI or escalated, not really about simple transaction anymore. And they are moments they're about moments that really require emotional intelligence and brand alignment. And this is reshaping support roles because training is different. It's not so much, anymore about speed and process, but really about empathy, decision making, embodying brand values in every interaction. And I think this is the biggest win, for ecommerce companies with the proper setup because you can basically use really your human touch points to show what is the brand we're standing for. And we see that especially in, our clients in the luxury segment where where really I mean, the anxiousness for deploying technology that is not human controlled for luxury customers is very, very high. So shifting basically training on enforcing brand and values, it's really, really a great opportunity here, and we see really, an up an uprising in conversion rates and everything. That that's a fantastic point, and it's something I hadn't really thought as much about. But you the this kind of adherence to brand values, we had talked before about the kind of, like, AI being used internally in that Copilot. Like, that's a a great way to be able to reinforce that training around brand values and your tone of voice and how you interact with people. But it's also when when people have more time back, it it actually gives a lot of these more luxury brands the ability to provide that white glove support treatment to even more customers than they would have been able to do before. Exactly. Yep. And, Chip, you know, not every conversation right now is being handled by AI at the moment. You know, maybe maybe in the future, it it it will be, but we know that this process of escalating a conversation or a ticket to a human is very important. So when people are thinking about that, what what would you say are some of the most important things to consider or to include during that kind of ex escalation process? Yeah. I think I mean, the word here is context. Like, when, somebody gets that support taken when there is an escalation, humans need really the, the session story, not just the chat text. They need to understand what happened, what led to that moment. Having that, you know, full picture, really gives them everything they need to really understand, and have empathy. So the customer was comparing these products. They tried this, you know, a particular size. They hid an out of stock. Maybe they had some kind of error issue with, you know, a promotional code. Giving that agent instant context and clarity, without having to go back and forth, you know, reduces handle time. It gives, like, higher satisfaction, unless you, you know, get to the resolution faster and create happier customers. And, you know, another, analogy we like to kinda throw around internally, you know, they've got the self driving cars now. Right? So you've got Waymo. You've got this idea of LIDAR that's, like, always on, always watching. You know, with companies like Fullstory where we can capture all of this behavioral data at scale, you really are capturing all of the unknown unknowns as well. It's not just what you tag or what you instrument. It's really a complete picture of that customer experience. And packaging that up and using AI to summarize that and then deliver that, whether it's to a human or or to a fin, gives it that kind of LIDAR that it needs to be able to understand and map and then know in real time what's happening. I think that's super valuable. It's all about the context. Yeah. Absolutely. Very, very well, Paul. Love that love that analogy, LIDAR. Great. And and, Tim, you know, so once an agent gets that context, how do you kind of envision them them using it? You know, like Chip was just talking about, if you have this beautiful picture of, like, what they've been comparing or what they tried to do, the the steps they've taken. How do you kind of envision the actual human component, taking that and leveraging it to to provide that that, great support? I mean, I don't really have to envision it because we are already having it in place. And, I mean, Chip is totally right. I mean, context really is everything that the agents need. Right? We work with in our own products, we put something that is called conversational assist. So, I mean, we we like to talk only about written volumes. Right? Because but most of the European markets are very phone driven still. So, if you have the right content, you have the right customer history and everything that happened before. We deployed conversational assist to basically nudge agent behavior in real time to support through conversational signals that are observed on the phone, like customer is at the risk of dropping off. There is an upsell opportunity. We should shift the conversation in this direction. So we are triggering alerts and coaching in real time. And those copilots help with conversation direction. Obviously, they improve coaching. And the result is pretty simple. So the agents spend less time guessing and more time actually converting and retaining customers, and it's something that every company should have a look into. Yeah. And, Ruth, I'd love to hear your thoughts on this as well. As somebody who's kind of at the the the forefront of this and and, you know, managing teams of people doing this, what are some of the things that you've seen or some of the trends or insights that that you've kind of gleaned from this kind of hand off process with AI? Yeah. We're doing something similar with our human support team now around giving them the context that they need as humans to see, how successful are our our customers being with our product. So we're able to empower them with context that they need, such as like how much is the customer using these like core features that would make them successful or these features that we know that if they use them they're less likely to churn. And the support team can actually go and look at that stuff now and action it with the customer. So you know they'll wrap up whatever reactive inbound conversation they're having and they're able to, with this context given to them, say hey I can see that you're not using x y and z would you like to have a chat about that? And Finn resolving so many of our more simple queries and a lot of our complex queries now as well has actually freed them up to have the time and space to do that with customers. That was a pipe dream a couple years ago when all we had was a burning queue and inbound work you just have to get through it and get on to the next thing Whereas now we're actually saying to people like, no. Stop and spend the time with customers and actually do this. And of course then it's like AI powered context and content, for them that they can use to do this with customers. And now we're working on different ways of giving Finn that kind of context as well. So Finn can have that conversation after Finn resolves it. Right? Then then Finn can say, hey. I see that you're not using this feature or this feature. And again, this comes back to the support team as a whole being a value add to the business, driving product adoption, driving long term success and expansion of customers rather than, like, answer the question and be a cost to the business overall. Yeah. I think there's a really good bit of advice in there because when we when we talk about people getting started even with repetitive queries, the advice is typically, well, look at what most of your common conversation volume is and you kind of start there. But what you're doing is you're taking the kind of real world picture of what's actually happening on the ground, and then you are then layering an AI to be able to to, like, solve that problem. And this is something very similar. You get to that point where now support reps are able to take certain actions. And then once you understand that flow, what is working, then you can, again, bring in AI to help start to automate some of that work as well. And I think that's another thing that's been contributing to a lot of the blurring of of different types of roles. I know we've had a lot of conversations with Junan on on the Intercom side heading up customer success where, this kind of the the clear delineation that used to exist between customer success or or, like, inside sales reps, for, like, these upsell opportunities, it's now starting to get blurred with the the role of support. So is that is that something that we're starting to see internally or have you seen with customers as well? Absolutely. Yeah. We don't really have a tier one level support position anymore. Like, the folks, that we had in those roles are after upscaling either to, like, more consultative type positions, and and doing deep troubleshooting with our customers that Finn isn't quite able to do at the moment, but we're still gonna figure it out. We have a goal of a 95% automation rate on our team. So we're living that much. Yeah. We're at 80 right now, so shout that out as a brag. But, yeah, the the profile of the folks working on the team does look different, you know, and when they have interactions with customers, they're trained up now. Again, not to, like, answer the question and move on to the next thing which is a classic customer support position and instead really spend time with customers going like explaining what went wrong in the first place explaining them how they can be, more successful with whatever they're doing next time around. And then a bunch of the the folks are also moving into support engineering positions which is a mix of answering the most complex and technical queries for customers but then also doing all this other work that's going to help our customers longer term so it's like off the queues and doing things like building tools for our customers or our team to use. So it's just this far more broad and honestly rewarding work that the support folks are doing across the board and then there's my team which is a whole new team within our company. I have a team of full time folks who are implementing AI functionality for our own customer support team. I have conversation designers, knowledge managers and some of them were frontline folks that I promoted into these positions because again they have all the context of our customers, how our products works and but now they're also upskilling so much in the AI space. They're able to move into these new, kinda cool and flashy roles actually. And support is really a team now where you can have, like, a much, a much longer career ladder than previously. So the whole the whole makeup of our team is starting to look really different to a few years ago. That's so exciting to hear and especially around the the ability for, further and deeper career development for folks. I think it's such a it's a great, kind of, like, positive externality coming out of a lot of this. Great. So let's talk a little bit more about some practical getting started and and kind of scaling advice. I know there's there's probably some folks listening, tuning in that are thinking like, yeah. This sounds great, but we don't have the time. Right now, it's gonna take too long. We have we have too many things in our plate. But, good news is you absolutely can be up and running with an AI agent in days. It doesn't take months or quarters to implement anymore. So, Ruth, I'm gonna put you a bit on the spot. So let's just say, you know, you've got one week and you have to launch an AI agent for, you know, like, a ecommerce brand or retail brand. Where would you start? Content. Content. Content. Content. Beef up the health center. Make sure it's actually accurate. Take the like, if you need to get some human bodies over to all the content or build out new content, or figure out, like, what are the most recent f f, FAQs. Sorry. I'm saying f four qs. That's another we we say feature request. All my acronyms. All your FAQs, like, just get people if you'd, like, take a hit somewhere else. If you're saying you're too busy or the queue is on fire, the only way you're ever gonna get ahead of that is, like, stop being so reactive for a moment and do something proactive. And the biggest piece with an AI agent is get your content in order and then Fin will start answering a huge portion of your conversations immediately. It really it actually doesn't take so long as well. You know, you might need to put some work and time into the content side of things, but FINRA will start resolving those types of questions the minute you let it ingest that information. So, what I would just say to anybody being like, I still can't get going with it. I can't actually prioritize it. At some point you have to be like, okay, I have to stop what I'm doing, this constant reacting on fire mode that I'm in and start making an iterative plan. And the other piece as well is that you don't have to go all in and have everything perfect with your AI strategy immediately. Like you heard what I was saying there that we started so small a few years ago where we just had the folks in in our inbox just using some basic AI features and before we went broad and launched out to our customers. So you can take steps and iterate and like I said now we're at an 80%, like 80% of our volume is being resolved by Fin. Like, we didn't start with that, you know, but we moved pretty quickly up towards that because we just kept learning and iterating on top of it. But content is key to the initial huge portion of success. So we just, like, get in there and make that great and Fin will start performing many for you from there on. Yeah. And would you would you advise people, like, you know, when if people are able to start from kind of square one, are there any other things they could be doing to potentially set themselves up for success later when they do want to start going into more kind of complex queries or complex use cases for AI? Yeah. I think what's cool for folks now is that the technology that compared to when when I got going with it because it was, nearly three years ago now, Now the technology is so much further along, there's so much more that AI agents can do than the initial informational queries from like when I got started. So get going with all that information and content but also start thinking about like what do you want Fin to do beyond informational knowledge based queries. What systems do you use? What does your team do? Like do you have to access Stripe or Shopify to do whatever it is that you're doing? And figure out those systems and figure out if Fin can work with them and those two I just mentioned yes they can and and it can actually like more and more easily over time as well connect to those systems. We have MCP that makes that a lot easier. Our partners here have all these different ways that they connect across the board with different systems so like the sky is the limit now at the moment so of course, while getting started, with the content is kind of the easy thing to do. Do start doing your research about what you would like Fin to do so you can start making a plan to get in that direction because it actually won't take you that long to get going with that anymore. Love that. And and, Taylor, I'd love to hear from you. In these kind of early stages, getting ready to start working effectively with AI in those earlier stages, how should leaders be preparing their teams to to be successful? I mean, it's not only about preparing, your teams. It's about preparing your company. Right? I mean, Ruth is completely right. I mean, if there was definitely an opportunity to start immediately and get going on, as you said, those low hanging fruits. But, I mean, we've see we are seeing a lot of failed pilots out there. And this is usually because data, wasn't ready. Goals were unaligned or teams were not prepared. And I think key for leaders is to set expectation and clear goals before launching anything because the amount of stakeholders you usually, at a scale up or corporate level, have to involve when getting AI ready is ridiculous. And you talk to legal, you talk to compliance, you talk to IT, you talk to your boss, your boss's boss, and then the board, and you get approval from everyone. So treat AI. I mean, do your pilots, build the use case, but then treat AI like transformation and not a quick add on because objectives, preparing your people, and also your culture to really embrace AI needs to be ready. Otherwise, you will never get into that agendic state where every everything goes hand in hand. I love that you touched on the cultural aspect because I don't think people necessarily understand. I think people will hear that and they'll nod their heads and agree, but I don't think people sometimes understand how big of a cultural and behavioral shift it can be for individuals and teams and even leaders. And I think that's a really important thing for people to be thinking about. This is not a, like, a quick project that's going to be one and done. It's going to be a complete reshaping of how orgs and teams and companies function. Chip, I would love to hear from your perspective. So let's say, you know, somebody's the they they've gotten started with AI. Maybe they're they're starting to see some of that early success, and then they want to begin to maybe, like, scale a little bit. I would love to hear from you what are some things they should be thinking about from that perspective. Yeah. I think it's, you know, in the same way that you need to be focused on content, I think you also need to be focused upstream on data quality and comprehensiveness. So it's a lot of times, you just don't have a clear view of exactly what your customers are experiencing because you're only measuring the things that you think the instrument or to measure. So, yes, you might know the happy path of how people move kind of down your funnel, but you don't know where the journeys are going sideways. And so having a, you know, having a, a set of data and a set of tools that allow you to really understand that customer journey, to understand where customers are hesitating, to, know where they're experiencing friction. That's so important to be able to then, to think about the tactics where you can actually implement AI to improve that customer experience. Because if you don't know where you're gonna have the biggest impact, your experiments just might not, you know, get the outcomes that you're looking for. And so get the right data, understand where your customers are are really struggling, and then think about the the customer journeys that you want them to go on, and then that's where you can deploy and have the biggest impact. Absolutely love that. And and, Ruth, over to you. So, again, like, as as they start to see success and grow, do you have any kind of, like, tips or or guidance for how they could be optimizing performance over time? I was just having this conversation with my team earlier on about, like, the art of the possible. And I know that's not like a like a hard like, something you can just go and deploy right now. But if you told me two years ago that we were gonna be resolving 80% of our inbound volume, like, we have a pretty complex product. Right? Fin is resolving that much. And we started out, like I said, like, with the the kind of simpler things that we built on that. But there's there's stuff that Finn is doing today that we we said not that long ago, like, that's not gonna be possible for years, you know. But it is you need to come out of that mindset of, like, an AI agent can't do something. Because even if it can't do something today, it's gonna be able to do it pretty fast. This technology is moving so so quickly. And now in this space where, like, I'm challenging my team to go from 80% to 95% in terms of our inbound volume being resolved by Finn, it's like like what can't it do and then what like document what you would need it to do to be able to resolve those questions. And then as the technology continues to develop or your team are really smart, they figure out workarounds, you can actually start to chip away at those remaining things that Finn can't resolve today or your AI agent can't resolve today. So, like, a tip would be, like like, think of it from the the positive and optimistic standpoint of, like, it is gonna be able to do that. Maybe it can even do it today. Because I definitely, I think it can be human nature sometimes where you're like, oh, I can't do it. Can't be bothered. Like, looking at that right now, no. It's not gonna work. And actually be like, well, it is. Like, even if not today, it's going to later. So let's start planning for that. And our future is, like, probably a lot sooner than we think. Yeah. That's a great point. The the future is closer than you think. That's one of the the things that's continued to ring true is the speed and pace of innovation in this space is only gonna continue to accelerate. Awesome. So we're getting ready to go into some of our final sections. I am gonna, just share up the results of that second poll. This, I was really interested to see. So we're asking what support requests or tasks, are you already already handling with AI? And there's a couple people people that said that they they aren't, but it see it seems like kind of a a pretty common mix of, the presale questions, you know, WISMO questions, and then also a good number of folks talking about complex tasks and then updating or changing orders, delivery updates, delay notifications. It it's great to see that kind of breadth breadth and depth, and I know it it varies from company to company, but I think it's a very good illustration that there's, so many ways to be applying an AI agent to this to this kind of space, whether you're just getting started or whether you're starting to scale. So it's really exciting to to see. Final reminder, feel free to drop questions, by the way, into that q and a tab. But once we get through this final section here, we're gonna have some time, for those, to go through live. And as we get ready to go into that, I wanna look ahead a little bit. So, Tim, I wanna kinda start with you. What do you picture as the maybe a little bit longer term or I I don't wanna say even long term because we know how fast things are happening. But what does just the future of AI in ecommerce and retail look like to you? I mean, I think it's gonna be fundamentally different and from what we experienced today. Right? And we as leaders, brands, also customers have to get ready for it. I mean, when we all started using ShareGPT, it was fun. We used it like Google, asking question, being amazed by the results, and some of us started processing documents. Now most of us probably already planning their travels with it or comparing products. So I think that if if technology providers or AI providers ultimately correct the operating system of your smartphones, customers will probably not visit brand websites anymore. Instead, they'll outsource basically their shopping experience to their own AI agents because they have all the information. They know the customer. Right? And we have a situation where AI agent speaks to Fin. So we have basically, an AI agent conversation without human interaction. And I think the big question is what happens when your customer does not land on your website and in your shop anymore? Do we need your shop? How do you get your shop ready for it? I think, Mark, we talked about something, how SEO is evolving to AIO. So, how do you structure for this future scenario? Because, ultimately, I think this is how I wanna shop in the future. Yeah. I I love this, and and it's something that I think a lot of people have maybe if you're if you're kind of deeper into the AI space, you'll start to hear about this concept of personalized AI agents to do any number of things, managing your calendar, your email, being able to kind of build these on your own. But the the concept of an AI shopping assistant is not all that different. I mean, people some people, especially if they if they're, you know, busy or if they have the the time or the money or resources, they will hire personal shopping assistants. Right? So this concept of outsourcing, that kind of, browsing or initial shopping is is not a a super foreign concept, but it it opens up a very interesting kind of can of worms. We're starting to think about what happens when somebody might be buying a product or wanting to buy a product, but they're never landing on your actual brand website. I think that's a fascinating thing to be thinking about. And I'm not sure if Chip or or Ruth, y'all have any any kind of, like, thoughts or opinions on on that. But if so, I'd love to to hear them. Yeah. I'll I'll take a quick swing. So just while, you know, as a AI user myself, and I love the vision for that, it it'll be an interesting period of time from here to there. And what I mean by that is brands rightfully will fight for their own experiences, for their own digital, you know, surface area. And I think that's a good thing. Like, AI systems are generalist, whereas every brand can really deeply own, the things that they, you know, that they're specialized in. And I think what it's gonna force is that, you know, these digital retail experiences are gonna have to evolve rapidly from kind of very traditional experiences to ones that feel much more personal, to feel much more connected, and, you know, a place where you want to be. And so while I think the channels will become more distributed more and more even as they are today, I still think there's gonna be a it'll be a road, right, to get from here to there, and brands will need to figure out what it means to have an experience that really compels users to to show up and to and to shop. Yeah. Okay. Ruth, anything that you wanna add to that? It this just reminded me of a story. My teammate told me recently that he wanted to buy a secondhand car, a specific type of secondhand car. And, you know, he was, like, trolling all the different secondhand car websites. And he was, like, why am I doing this every day? Like, you know, and he might set up some manual notifications, but he just put a prompt into chat g p t to troll those websites for him and notify him every time a car of that make and model and year, that came up for sale. And that immediately meant that he wasn't going on to those individual branded websites and, like, you know, it it was like Tim was saying, people are gonna figure out ways of of doing this and people are gonna, commodify giving you a way to do this. So brands are gonna have to come up with solid ways to be able to pull people in and make sure that, like, their experience actually needs to be better than what someone can get elsewhere as a third party. So, yeah, it's really interesting worlds that we're in. Just the amount of things that we can, like, ask AI to do without having to, like, go to these, traditional spaces. So, yeah, all brands out there that are listening. I think everybody needs to, like, up their game in terms of their differentiation for why people should be, like, on your site and using your AI solutions to shop with you, because people just want it to be really easy and really smooth. Absolutely. I love it. I've got one last kind of rapid fire question for everybody. So maybe try and let's try and give us, like, a like a sentence or two, and then we'll go into the the the live q and a. Ruth, we'll start with you. What's one piece of advice you'd give a leader about using AI in the retail and ecommerce space today? If you're not doing it already, you really better get going because this is a competitive space, and you're, like, you're going to be you're gonna be, outperformed by competitors. If you're giving if they're giving customers a seamless instant answers to whatever they need, good nudges and proactive, you know, that's where they're gonna go and they're gonna leave the space where they wait forever for support queries to be answered or they get really annoying proactive pushes on their on that website, you know. So, you need to, like, get moving quickly and make sure that it's really good and differentiated. Love it. Chip, how about you? Yeah. I'd say every single conversation should be an opportunity or a signal to improve customer journey. It's not just a ticket to close or deflect. Like, learn from those conversations. Absolutely. And, Tim, how about you? Yeah. I mean, to build up a bit, get your house in order, align your system and teams, and then move fast people and scale with confidence. Love that. Thank you. So, I'm gonna go ahead and, pull up some of the questions that we have here. I see one from Tom Hopkins. I will share this up and read it. Thinking about the eighty percent resolve rate from Ruth, what's the definition you're using? They wanna think of how how we're comparing it. Interaction AI did not resolve a contact click. So, basically, like, what is the definition that's used for, like, an actual resolution, for for a conversation? So how we officially measure it, within in our in our product is, something called involvement rate. So of all the conversations or emails that you get into our Intercom platform or if you're using Fin, any inbound interactions that you have, how many of those are you allowing Fin to be involved in? And then of that number, how many are actually being resolved by Fin? So that's the resolution right side of things. But I touched on something that we're starting to speak about a bit more now called automation rate. It's actually the broader picture. So of all your inbound contacts, how much is your AI agent actually resolving? So it's like of everything. And for us at the moment, it's 80%, which I I'm pretty proud of. Love it. I got another one here. Rajat, I hope I'm pronouncing that right. For a small company that's just starting off with the customer success and support, is it meaningful to practice with AI agents, or would you suggest going through that manual cycle first before implementing an AI agent? Maybe, Tim, you wanna, like, start with this one and then, Ruth, if you have anything to add. Both, I would say. Practice with AI agents if you have time as you can free up someone on your team because those learnings are invaluable, but also do the manual cycle because doing the manual cycle means learning about your product, about your customer journey, and about every touch point. So, unfortunately, Rajesh, you have to do both. Great. There's a pretty interesting one here from from Tom, Hopkins again. And this is this I'm wondering if this is gonna go over my head, but, there's a he's talking about a lot of the kind of, like, research in the mathematics around the ability for models embedded models to get very accurate responses. Details. So for any space where accuracy is paramount, how do we think about the new work of human and where humans need to fit in before it responds to to the customer? Open to anybody who wants to maybe, chime in on that. Chip, you wanna stop? Yeah. So a lot of data scientists and definitely not reading these papers. So it's a great question. I mean, I'll give you my perspective on how we're thinking about it at FullStory. So one of the things that we can uniquely do is provide, like, real time contextual data about what the user experienced leading up to a conversation. That's kinda how FullStory and intercom fin work together. What we're doing is trying to create systems within FullStory that allow for that human eval loop to make better and better summarization of that digital experience. It will improve the accuracy of Fin's context window effectively. And then the closed loop system that I think gets really exciting in some of our kind of frontier research is as conversations, can be fed back into success of those conversations back into the digital experience analytics. Like, you can really understand outcomes broader than just the conversation. So think about deploying kinda, more up funnel, experiences that are gonna drive conversion rate. Well, how effective are those, and can you create kind of feedback loops that allow the AI to get, or at least our part in the AI to get smarter, to improve, to drive better outcomes. Love it. Thanks for thanks for, coming in coming in with that response. I was definitely not gonna be able to to answer that one. Awesome. So we're we're about out of time. Just two very quick announcements before we wrap. First, not sure if people know, but Intercom just became officially a Shopify Plus partner. And as part of that, we released some really amazing new updates to our fin and Shopify integration. You can manage multiple storefronts, handle refunds and order updates automatically, keeping data in sync. So, in that doc section, I linked to the integration. You wanna check that out if you're a Shopify user. And, also, I pinned it in the chat here, Intercom's annual summit for AI customer service leaders pioneer. It's happening on October 9. You can click that link to register to attend the virtual livestream that we're gonna be doing, or if you're gonna be in New York, you can even look to attend in person. Outside of that, in that doc section, please go ahead and take a take a minute to check out the The Nest by Concentrix. We've got a few really cool we have a a podcast from Concentrix in there you should probably check out, links to their site. Same thing with Fullstory. Give them a look. Check out our integration that we have there. And, outside of that, just wanna say thank you again to everyone for coming and joining us today, but a special shout out to our, amazing panel for taking the time to share their insights with us today. Thanks, everybody. Bye. Thanks for having us. Thanks so much, everybody. Thank you, everybody.