Building the Future of Customer Experience with Alan Ranger of Cognigy

So what the consumer does is they they send in a photograph of the thing they want to return. So it might be a jacket with a damaged zipper and then the AI will look at it and work out whether it was damaged because somebody forced it or whether it was a faulty product. Then if it can't make up its mind, it will then go and consult a human. So it can decide that I'll go and ask a human and it'll send the team's message to a human with the picture and say this is what I think it is. What do you think human? And the human will then respond back through to the AI agent who will then go back to the consumer. So yeah, you got that reasoning and humanlike sort of capability in the product itself. But to be honest, we're we're not

really doing it within the the business itself. I don't think we're quite at the scale and obviously we're really focused on providing stuff for our customers. >> Hi, my name is Demetri Bonichi and I'm a content creator, agency owner, and AI enthusiast. You're listening to the AI agents podcast brought to you by Jot Form and featuring our very own CEO and founder, Idkin Tank. This is the show where artificial intelligence meets innovation, productivity, and the tools shaping the future of work. Enjoy the show. Hello everyone. My name is Dmitri and in this episode we're talking to Alan Ranger from Cognagi. How you doing Alan? >> I'm really good. Uh delighted to be here. So thanks for having me. >> Absolutely. Yeah. So I'm really excited to chat with you about Cogni. It's uh obviously you're the VP of marketing at Cognagi and you've been there

for seems like the last uh you know three yearsish, right? >> Two and a half three years. Yes. Yes. It's probably the most exciting two and a half, three years of my life, I have to say. >> Okay, tell us a little bit about that. >> Yeah, sure. So, um, if you could see me on video, you'd see I have lots of gray hair. I've been around a very long time. So I first started um in this space in about 2000 in fact no 1993 selling the first software ACDs that didn't really work so for call distribution and then sort of worked at places like Oracle CIA and big enterprise but really got into uh the contact space about uh 15 years ago and pretty much stayed ever since. So um yeah I've been at Cogni for the last two and a half years and

as I said it's been tremendously exciting the the pace of change and the the revolution of AI uh within the contact center has has really changed everything. So really excited about what's next. >> Awesome. Yeah. Well, um I guess how did you get into working with Cogni in the first place? Kind of like what what was the the leadup personally for you, you know, kind of getting into this uh position? >> So uh before Cogni, I spent six years at live person working with Rob Picassio and uh one of the guys that I worked with there had joined Cogni to lead sales worldwide and he called me up and said, "Hey, there's this fantastic company in Germany. You have to come and work." and I met the founders and really liked them, really loved their vision of what they were doing. Uh came to the

offices, met all the people developing the product, experienced the product and just thought yes, this really is the next big thing. Uh it's too good an opportunity to pass up. So yeah, joined the company uh for the product and the founders and I've stayed and really enjoyed it because of the amazing team that I work with. >> Awesome. Very cool. Um and I guess could you just let everyone know a little bit more about um what Cogni does? obviously uh you know you're in the uh AI agent sort of like customer support space but we I'd really want to hear kind of like from from you what uh you think is you know your core value that you provide to the market right now. >> Sure. So Cogni exists because unfortunately customer service still sucks. There's an awful lot of companies that are still using

really antiquated old bots that just don't understand and they're frustrating. It's almost like they're sticking to the strategy of we don't want to talk to our customer, let's find a way of deflecting them or, you know, containing them and and not actually having a conversation. So, what we're doing is we're we're revolutionizing the whole sort of customer service. It's it's not just inbound, it's also um outbound and uh I'll get into that a bit later, but yeah, we we make AI agents and they're just the same as your human advisers that you have uh within the contact center. Uh people need us because there's a huge shortage of human advisers in the world. Uh particularly after the pandemic, nobody wanted to go back to the contact center. They all found different careers and jobs and that sort of stuff. So everybody that we work with

all universally say it's so difficult to obtain and train and retain good human advisers. So what they're doing is looking to fill the gap that's happened now in customer service. I mean customer experience has got worse since the pandemic. You know whole times have gone through the roof. Um the technology that's been deployed people just haven't adopted it in in full yet. So there's a lot of lot of stuff to be done within the market to improve it. And these AI agents can take over typically to start with some of the simple tasks that humans shouldn't have to do. So the routine tasks, the ones that need to be done but don't necessarily create a high value. And that leads the human advisers to focus on being sort of providers of value and having the relationship with the customers. So in no way are we

advocating let's get rid of all the humans and replace all AI. Let's just make sure that the humans that do work in the contact center become value workers and actually get some fulfillment out of their jobs because it's pretty miserable. I wouldn't want to sit there changing passwords uh all day every day. It's just not needed for for a human to do it. So the way that we do it, we work with all of the big CCAS vendors, you know, everybody from Genesis to Nice to uh to A Via and the newer ones like Zoom um and uh uh you all of the other ones. Uh and what we do is it's very much like having a pre-trained AI agent or human adviser coming into your contact center. Uh you can sort of start off perhaps with one that um is your concierge. So, effectively

replacing the old IVR. Um, because of what we've done with Aentic AI, and I'm sure that's something we'll get on to in just a second. Uh, the quality is so good, it can recognize pretty much everything being spoken to it because we've got the large language model behind it. So, it can the um the accuracy rates of understanding intent and being able to route somebody to the right uh place is amazing. We've got one of the largest insurers in Europe. Pretty much all of their calls are now answered by their AI. Uh and that AI what it does is it first of all um does the identity and verification checks. Uh then it does the uh intent analysis. It works out exactly why the person's calling. And then what it will do then is give a warm hand over to a human adviser. So the

human adviser knows that it's Dimmitri and he's calling about the the car wreck he had. Uh and he also knows that the the human adviser will know that you're asking do you get a rental car as part of your policy. And then at that point it changes um roles and becomes a co-pilot and works in the background assisting the human adviser coming up with next best responses. If it's in a regulated world like insurance um it will make sure that certain statements are read that sort of stuff and then at the end of the call it does the call back up summarization and then we'll go and do the sort of the update to the backend system. So yeah that's a very popular use of conversational AI and it just improves the customer service fantastically but also takes probably 25 to 30% out of every

call that a human normally would have to do. Okay. Wow, that's a lot of lot of percent. Um, excuse me. So, what I guess I would say is with this tool, if I understood it correctly, is this something that you implement directly into uh, you know, you said things like Zoom and whatnot? I think some of the different integrations that you have like are you are you working with the those companies themselves as like integration pieces as like a vendor so to speak or um I guess how do how do those relationships work specifically? >> So some of some of the some of the CS vendors do we resell us as the uh an OEM was the AI agents but we work as a standalone platform. So the the AI agents are created using our our platform. Uh it's really quick and easy to create

AI agents now and I'm sure we'll get into that in just a minute. And yeah, we work with all and it's all done with connectors. So yeah, we're able to connect into pretty much every major platform. We have customers across the whole spectrum. So yeah, it's it's a wellproven technology. >> Okay. Um, and you guys have been doing this for I want to go back a little bit, right? So how how long did it take for Cognagi to get to this point, right? because I I think it's a fair assessment that we've seen some rapid adjustments in how you know AI agents have managed to improve and managed to get to a point to where they're feasible, right? When did we get from idea to feasible timelinewise? >> So, that's a really good question and really important for anybody that's looking at this because there,

as you say, there are so many new companies that have appeared that can do amazing demos of brilliant AI agents you can have a really good conversation with. Yeah, the company was founded eight years ago and that was about the time when conversational AI first came along. Before that, it was just sort of um the the old bots that didn't really understand and the reason the company came into being was the the founders Phil and Sasha were looking around for a really good NLU to to build their platform on. they couldn't find one so they built the way and for the beginning and for probably the first six years of the company's life we just focused on making it easy to build really good deterministic um AI bots so you know being able to build stuff with flows make it really easy so you didn't

have to be a coder like most of the original platforms then yeah then everything changed when large language models appeared on the scene so that was about the time that I joined the company it was January um 2022 and obviously they're 23 three other obviously at the end of 22 chat GPT appeared on the market. So we joined and we took the decision that we were going to embrace the large language models and and we were one of the first to actually find some real commercial benefit out of it. So yeah it took um about six years building deterministic stuff to give us the experience of building stuff that would work at scale. So we have um one of our largest customers is Luanser. Um if they shut every airport in Germany because of a industrial action all of a sudden all of their flights

are being rebooked by our AI agent. And when this happened 6 8 weeks ago, uh it was rebooking and and taking 10,000 messages every minute. We think it's one of the busiest AI agents in the world. So it's building that stuff for scale, but also building the integration into the back end systems. That's the stuff that takes all the time. With the newentric technology, you can build a really good AI agent in a day. You pretty much just describe what you want it to do. You give it a name, you give it a role, you give it the tools it needs, you give it the guard rails that it's allowed, you give it the knowledge base that you want. press a button and you've already got one working ready to start testing. >> Interesting. Okay. So I guess you know it's pretty yeah it's pretty

common uh thread that I pull on with with a lot of the companies that we talk to that the transition really seem to occur and get expedited by the implementation of large language models and if you're open to it I I mean I don't want to get too technical But how did it specifically and does it specifically function? Right? Like obviously you can't really get into all the tech, but so to what level you're cap comfortable explaining cuz there's a lot of different ways that these these tools work, right? Like you can have a pretty good easy to use front-end experience where you would add knowledge bases and stuff like that. Then they use rag functionality. like what kind of level of detail are you are these tools getting into for like specific uh clients and and how does that necessarily work because I'm sure

people are mainly focused on the specificity of the um capabilities not general customer support right it's for specific companies >> yes so uh when we first started in 23 uh we thought that it would be mostly around testing so we were using the large language models to automate the testing of of AI and that was a very valuable use case then pretty much everybody including us jumped on the co-pilot uh and decided that the best thing for the large language model was to run in the background uh and provide support for the human advisers you know doing things like providing knowledge bases that sort of stuff but we took it a step further we um built it in as I was saying you know it's to part of the the conversation so it's listening in to the background of what's going on and then actually

providing the nesteps responses all that sort of thing but the real value is in consumerf facing uh AI agents that are powered by a large language model uh and the way that we're using them is that for the example of the um the concierge, the IVR replacement, the large language model is being used to understand absolutely everything that's being said, whether it's in a message or whether it's a voice call. Uh and that's working out what the intent is. Um and then it's um going probably in most cases still within the old sort of deterministic flow-based stuff. So an awful lot of companies are like, let's let it have a conversation to find out exactly what's happening, what the intent is. And then once we know what the intent is, if it is something like filling in uh an insurance claim, it's much better to

then drop into a deterministic flow, still using the large language model to craft the responses. So it's still a natural conversation backwards and forwards between the the human, sorry, between the AI agent and the the consumer. Uh and then at the end of it, yes, obviously being acting back in c-pilot mode, being able to understand everything, do the transcription, being able to do the updates, but a huge amount of value is now, as you were suggesting earlier, is in the creation of the AI agents as well. So, it's being able to the large language model to describe what the AI agent is going to do, give it the knowledge base, um give it the rules, give it its role, uh and just keep sort of iterating to to build it out so it does exactly what you want. >> Okay, that's that's good to know.

And and regarding voice, right, obviously it's uh it's verbal, correct? So it's on >> it's it's either voice or it's me digital. It doesn't matter. It's the same AI agents for both. >> Exactly. Yeah. And and my question is we've seen interesting improvements in a couple different models probably in the last three four months between like 11 Labs got like a big upgrade uh to well it's currently in alpha so that one probably doesn't count. Um but Chache BT had some improvements. Um and by the way I mean the V3 11 Labs alpha for those who are aware. Um Chache BT had some improvements on its voice. Are you seeing steady improvements on the voice side or is that proprietary to you using third party tools kind of how does that kind of fit in? Because I I'm sure that's one of the big concerns

right is that it sounds like a like a person. It's working really well and yeah being an open platform we use the best of breed so we do a lot of work with 11 labs and deep grab really the two that we work with and yeah the quality of the voice is tremendous the intonation you know the it sounds like it's empathetic it sounds like it's understanding of course it's still a cold-hearted machine but it sounds really good and people love it I mean and the reason um that we're able to now start doing outbound is because of the quality of that voice and the humanlike interaction because if you're just doing inbound customer service for a start people will put up with a bit of friction Uh, so if we were about to have this call and my broadband was broken, I'd have done

anything to get it fixed. I'd have gone on hold for 20 minutes. I'd have spoken to 10 different people. I'd gone back on hold. I'd have reidentified myself twice just to get it fixed. But if somebody was trying to sell me broadband with an automation and I had any friction at all, I'd just put the phone down. So the the the quality and the ability to have a conversation because when you're making an outbound call, you've got no idea what the person on the other end is going to say. It might not even be them that's picking up. So, we've now got our um our first sort of outbound sales and marketing uh AI agent live with a very large bank in Europe. And what they're doing is that um people when they purchase uh a loan to sort of buy a new car or

to renovate their house quite often will take out loan insurance just in case they lose their job and they can't make the payments, that sort of thing. So, what they're doing is that everybody that's bought um a loan and has ticked the box, yes, I'd like to be contacted about loan insurance. the outbound AI agent is making that outbound call. Beforehand, they had really expensive regulated financial advisers doing outbound calls and they had a 15% pickup rate. So, these guys are just dialing and not getting anything at all. Very expensive, getting frustrated. What we've found is that with the AI agent that then places the call, it's so natural and personalized to the person picking up at the other end. It's getting an 85% success rate. So, what it's doing is it's calling saying, "Hi, uh, this is name of the AI agent from the

bank." um calling about your application for for loan insurance is now a good time. It'll take about 10 minutes. If the person says no, it schedules a call, calls back. If the person says yes, it then does the warm handover to the expensive financial adviser and the financial adviserss love it because they're only talking to qualified people and yeah, it's made a huge difference to them. They're rolling it out across the the bank worldwide now and they're absolutely delighted with it. So yeah, that's a really good example of having a really good voice with really good agentic capabilities of understanding everything that's being said is dramatically going to change the whole customer experience, not just customer service. Yeah, it's it's really hard to like put into words unless you're in those sales cycles or those conversations how frustrating it can be when you're someone who

you know and obviously every everyone's time is valuable, but someone whose time is especially valuable and you're going into a call where they're they're not qualified and there was a disconnect and disconnect in some respect as to what value they thought they were bringing or providing. Like I've had sales calls recently where people essentially thought that they were like going to help me out with um creating content for like what I do at my agency, but I was under the assumption it was going to be a discovery call, right? So those these are these are the types of things that we want to get rid of for sure. Um it's it's very frustrating and you know I'm curious with this where you see adoption then so you're mentioning adoption a little bit right and and where you see adoption changing and moving forward like what

what have you seen adoption wise uh comfortability with like companies as to how much of a percentage that they're kind of taking your agents and putting them in and where maybe there are even more opportunities for for them to do it and not just like customer support but like I just mentioned sales too. >> Yeah, the the big area of adoption is obviously an inbound customer support because you do need to do that stuff, you know, making sure the right person is given the answers that they need in and a really good experience. So that's where it started. What we've been really surprised about is the rate of adoption of aentic AI. So you know moving from the old deterministic flow-based very reliable you know it's going to work um to the agentic stuff and brands are doing it just because they have to because

of customer experience they're being driven because their competitors have now started adopting it and there's all this pressure on the people in the contact center to now deliver you got the the chief executive is pushing down saying hey if I can have a conversation with chat GPT and just talk to it why why do our bots still suck you know you've got to do something and the consumers as well are now expecting as well so the poor people in the contact center in the middle are being squeezed from the squeezed from the bottom. We've also found this new role within many large enterprises of the vice president or senior director of AI transformation and they've been tasked by the board because probably McKenzie and the others have said hey you can improve your bottom line with generative AI and the first thing they go and

look at is customer service mostly because if you ask chat GPT what it's good at it says I'm really good at customer service so that's where they tend to start. So we we're really seeing that person as one of the personas that we're selling to and yeah, so there's all that pressure on and so that's really sort of pushed into customer service. But as I was saying, all sort of sales and marketing can now be transformed and we're beginning to see it happen. Um we're seeing uh everything from consumer goods companies that never had a relationship with their consumer before. So take shampoo for example. The shampoo company has no relationship with the person washing their hair at all. the person who's washing their hair buys the shampoo from the store or gets it delivered online or whatever. Uh but now if they want to

have have a question about it, they can scan a QR code on the back of the shampoo and it will start a WhatsApp conversation directly with the brand and the brand or the AI on the other end will know exactly which product the person's holding in their hand and can then start the conversation. It could be that, hey, I I used your shampoo and it made my hair go frizzy. And so immediately they know that the person that's calling has got frizzy hair uh and the shampoo that they're using didn't fix it. And they can have a conversation actually resolve their issue. But once that connection is in place a few weeks later, they can go back and go, "Hey, how's your hair? You know, we've got this brilliant new conditioner for frizzy hair." And then do the follow-up and then provide vouchers through the

channel. So all of a sudden you've got onetoone hyperpersonalized conversation. >> Yeah. hyperpersonalized like almost uh the drip campaigns that always have existed, you know, casually in marketing, but so more so in Yeah. Wow. That's actually impressive because they they've taken it from the analytic side of like, oh, we see some consumer behavior to, well, we had an actual conversation. Yeah, that's good. >> And the conversation persists unless the consumer chooses to end it. You know, it's like having a WhatsApp conversation with your friends. They don't ever go away. >> That's a good point. Yeah, because it's just like stored messages and cash and context. Um, so >> yeah, as a marketer, that's the thing that excites me the most, this capability of having one-to-one conversations with with millions of people. >> Yeah. I guess a quick question for you, just since you you know,

obviously the position you're in, marketing, what uh what have you noticed to be easy and maybe difficult, like the the good side and maybe the hard side to marketing a product such as this, right? like just from your perspective, >> we were probably the first um saying probably because I'm sure somebody else may have done it before to start saying it was but it was we were uh talking about AI agents. So this was to just after I joined actually about two years and a bit ago we started telling everybody that we make AI agents mostly to differentiate and give ourselves some space between the old bots and uh and what was happening now with AI agents. So we stopped saying it's all about building a conversational um AI platform da da da. was we will help you create AI agents that will supplement your

humans and that was our story. And what's become really easy now is because everybody has jumped on the AI agent bandwagon including Salesforce. They've done a brilliant job of educating the market with agent force what an AI agent is. So now we don't have to tell people what an AI agent is. What we have to tell people now is that um you need to do it with us because we've done it before. We've done it with some of the world's largest brands. You know we're rebooking 10,000 flights a minute with Luansza. So if you want anything that's enterprisegrade, you should really look at us. You know, we we're a German company, so we've gone through compliance with one of Europe's largest insurance companies. So we can get through GDPR and all of the other European privacy regulations. All that's good because as we were saying

at the beginning, there are so many new companies that are basically a large language model with a wrapper around it that can do a brilliant demo, but once you get to maybe the second or third meeting, you realize that they just haven't done it at scale with an enterprise before. So that that's the easy. The difficulty is there are so many new people and confusing the market. You know, uh there's some very high-profile startups um that have appeared and they do a really good job of, you know, getting engagement at the sea level and doing the demos and that sort of thing. And what has actually helped us, which we never thought it would, is the the rise of what they're calling the AI council within large enterprises. So this is all of the stakeholders within the business are now in the AI council and

nothing is bought or implemented without the blessing of the AI council. And what that has done has stopped you know chief executives and leadership appearing with a product that one of their friends showed them on Saturday night and saying we must implement this because it then goes through all of the rigor of enterprise systems. So the IT department takes a look, the compliance people take a look and we thought it was going to slow everything down but actually it's been in our favor because we can tick the boxes. >> Okay. So that's interesting. AI councils. I actually haven't heard this. I've heard some adjustments have happened in in industry in similar respects, but maybe not, you know, and at the enterprise level. And I've even heard that, you know, university level, they're they're considering offing a lot of their CS uh departments and sort of

moving towards more of an AI focused uh approach so that basically if kids are getting out of college, they university rather. Sorry, I know you're from the UK. I try to get the name right. uh uh uh you know it's interesting yeah at the enterprise level that would be the case. Um, are there any security or any like I I I mean obviously the GDPR thing so it's like that's got to be a common thread, right? Like uh I know consumer kind of data and all that kind of stuff is always something that people are freaked out by and like we as marketers hear that whole ad contextual selling based off conversations and other people freak out and I'm like but you want the conditioner, right? Like I I know you do. Like you want your hair to be better, right? like what's so bad

if I have it in this row in this database somewhere like you know >> yeah there is obviously huge focus on making sure that we're compliant with everything you know double opt-in all that sort of thing in terms of the the market detail but also on the data itself you know we have a lot of customers that really do need absolute cast iron guarantees on data sovereignty so you know nobody outside of the EU for a EU customer can have access to any of the data under any circumstances all hosted in the EU all that sort of stuff which you makes the the dream of really easy and um reasonably priced cloud deployments um a bit trickier but you know we were able to do that with with Amazon and with Microsoft so yeah a lot of our customers are using Azure and that sort

of stuff where it is guaranteed to be natively um within the country and then yeah it's all the privacy regulations everything from the EU AI act which insists that there's human oversight right the way through to GDPR and that sort of stuff but yeah I mean once you've been through it with the compliance it still takes a bit of time but uh yeah again when you've ticked the boxes a few times once you get used to the questions. >> Yeah, fair enough. Um, what are you most excited for uh in this industry moving forward? Obviously, if you want to talk more about that marketing piece, I think you could. Um, I I think there's a lot of really cool opportunities for uh improvements in in business in general. But from a from a marketing standpoint, what like what are you excited about? >> So, from

tell you, I'll give you a really wild future prediction. I think that we won't have anymore. I think that we won't have won't have websites. We will have a chat window and you'll interact with it in natural conversation and within the chat window all of the things that you would have found on a website will appear. So you could uh you know be so you could start the no reason why you shouldn't have a dynamic chat window that this fills the whole screen of your browser. So maybe you're trying to buy a car and uh it would you know you you'd have the conversation and it would be showing you all of the things you would normally be searching with Google to actually find it. So I think that's quite possible within the future that we won't have a browser as we know it interface

to get information. We'll have a conversation with a device with a screen that will show what we're talking about. >> Interesting. So like almost you're familiar with the new ICP type stuff, right? with um no or not why did I say ICPI sorry MCP I misspoke >> oh yeah MCP yes yeah >> interesting yeah because that's where that's where my head's at with all this conversation >> wild prediction yeah as I say what I'm really excited about is that step in between of being able to have all customer interactions handled through AI with the humans providing value where they need to >> h so for your company right is there any uh kind of interesting gleanings that you had along the way of whether it be marketing or building out the product that you'd recommend other companies trying to do maybe not in the same

exact AI uh support chat space but in other areas of AI. um anything specifically that like kind of was a was a roadblock that you felt like you could you could help somebody earlier stage uh understand how you know what what you know issues you faced and and how you overcome them how you overcame >> I think it's it's the it's just applying sort of common sense and you know focusing on the outcome so why are you in business keep asking that question why are we doing this as I said why to the beginning the only reason we're in business is because customer service still sucks um so you really have to focus on that and being German and being engineering when I joined the company. It was it still is a tremendously proud engineering company that builds a software platform. And what people would

do within probably 30 seconds of meeting everybody, they would whip their laptop out and start building flows and doing demos. Uh and that really isn't the where the value is. They what they should have been doing is as you say, have a discovery call and actually work out where the biggest pain points are. uh and then work together to actually resolve those pain points and then get the sign off that you will have the budget to get those pain points resolved and then actually go ahead do it as planned and being able to show measurable results. So we always recommend sort of start with something relatively simple but with a big impact and a big ROI. Uh so something like replacing the IVR with a conversational concier or maybe just focusing on one thing like uh Lufansa did to start with which was rebooking flights.

So something with a big impact that you know you can achieve and do it really really well rather than just having this conversational AI platform and trying to roll it out across the the company without a real strategy. I mean with this new role of uh sort of VP of transformation we have seen a few projects where people are just getting the platform and just playing with stuff but they haven't actually agreed with the business what the outcome should be. So yeah it's making sure that you focus on the outcome that's agreed with the business. Obviously the platform features and everything that comes further down the funnel in the sales cycle because that can be a differentiator. But don't you know try and start wowing people with uh with the platform. It's all about the experience and all about the outcome. >> Yeah, that's a

good that's a good point. Um I feel like that's actually just general good business practice, right? Like you you >> amazing how many engineering le businesses just want to show you what they've got. >> Yeah, that's that's a good point. You know, I I this has come up on a couple podcasts in a row uh where somebody acknowledged that there is sometimes a disconnect between uh engineering and and business in that sense. Like for example, uh you could have an engineer for a project management system not understand why if you tell them, could you put a check mark here on the screen? And they're like, yeah, I could do that. But it doesn't quite mean anything unless the rest of the system is fulfilled in such a way that the check mark actually does something to uh move the product forward. So obviously we've talked

more about customer support and those types of agents but for other agents for example chatbt just released that recent improvement that I'm sure you saw where there was uh agent capabilities. What other cool aspects of the AI agent suite workforce so to speak have as a company you guys implemented because obviously this is like two different things like you are actually doing a specific type of AI agent but where in your own workplace have you managed to implement AI agent? >> Yes. So we have actually used our own technology to support our customers. You won't be surprised to hear. So that's an obvious one but we did. Yes. So yeah, so so we do that but also yeah internally within my marketing team um you know we're using we've got our own um trained models to which should be trained in our language and our

tone of voice. So we've got everything from a brand checker where you can just sort of put in what do are these words on brand? You know you can just write something or probably using a large language one will help you write it. Put it in and it will then come back and say no that's not on brand this is good these are the improvements right the way through to being uh able to respond to RFPs and RFQS. We've done it so many times, we've been able to train a model to be able to actually do 80 to 90% of all of those responses before it's checked. Uh, and then on the other side of things, with the new capabilities of chat GPT, particularly, we're now um encouraging all of our sellers to pretty much rehearse their next sales call. So, they'll say to chat

GPT, right, you are the IT director of a large pharmaceuticals company. You're in a bad mood and um, this is the use case that I've come to talk to you about. and they actually then practice their pitch and they'll have a natural conversation with CH GPT to actually do that and that's had a great result. It's it's really highlighted um where sales salespeople need to uh hone their skills and and make adjustments to to the way they pitch to people. >> Interesting. Yeah. So, have you managed to do that with the voice version? Um or is it more Yeah, the voice version. Yeah, cuz I actually learned that a little while ago where I was basically like, "All right, be the most irritating, impossible to sell person, right?" And Yeah. Yeah. It's It's at scale, too. It's like you don't have to compete. Yeah. It's

very It's I I think I think that's something that's incredible. Um, has the company taken advantage of any of the advances in AI agents in regards to what was previously understood as automations? cuz basically that that's been like kind of a revolution right with MCP everything like that you know it's it's moving out of you know make.com and you're you're really good still but more highlevel agent integrated API protocols you know like cuz it's a it's it's something that if then logic went from being the only way to really save a bunch of time to now have the agent think about about it through some if then logic and it comes to its own conclusions. >> We haven't really done that internally, but obviously within our own uh the agentic AI version of the the AI agents, yes, they're able to think and reason and

act. Uh so good for example, we have a fashion brand that does all of their returns now using an agentic AI agent. with a damaged zipper. Uh, and then the AI will look at it and work out whether it was damaged because somebody forced it or whether it was a faulty product. Then if it can't make up its mind, it will then go and consult a human. So it can decide that I'll go and ask a human and it'll send the team's message to a human with a picture and say, "This is what I think it is. What do you think human?" And the human will then respond back through to the AI agent who will then go back to the consumer. So yeah, you got that reasoning and humanlike sort of capability in the product itself. But to be honest, we're we're not really

doing it within the the business itself. I don't think we're quite at the scale and obviously we're really focused on providing stuff for our customers. >> Okay. Interesting. Um yeah, I definitely would recommend getting into something like that. It's it's it's one of the more interesting aspects of business recently. Um to the I don't want to say the few of us, but I think that would be an accurate statement. I'm just trying to not sound like greater than thou by being one of the few people. the there's a good chunk of very technical companies right now that are making really good AI agent stuff obviously like you're doing and there's also other companies that are just utilizing that aspect where it's like okay there's a lot of work from a foundational associate level similar to how the front-end call center people are at that level

and knowledge work that's getting automated and I always try to ask people who come on is that something that's improving in your business because it's definitely a big push like I've had the same question I posit to guess which is when do you think we're going to see massive labor market shifts you know for example I have this theory that got pushed back from 5 years to 3 years and I'm not sure where it's at now maybe it's at like 2 years I think associate level knowledge work jobs you when it's more so basic clicking around and and logic will kind of be out of the not out of the market but be starting to get phased out in about two years. What do you think about that whole impact of AI uh in the next couple years? >> You will because because once somebody

else is doing it they've got the efficiencies the companies that don't will just not be able to have the efficiencies and they won't be competitive in the market and they won't be profitable and until the turn they do. That's everybody everybody's going to do it. Um what we're seeing in the contact center at the moment, we don't have a single customer that's actually laid a human advisor off yet. They're still really busy trying to fill the gap of not having enough. Um so maybe that'll happen. Maybe there'll be attrition through natural wastage. Um but one of the really interesting things we've done um did it a couple of months ago in the UK with the CCW exchange. Uh we had a panel of young um uh call center advisers came and joined us. We spent a day with them showing them the capabilities of what

AI meant for them in their future. Uh, and by the end of it, they're really enthusiastic. And three of them really bravely got up with me on stage in front of 500 people, including their management, and just answered questions about it. And, and one of them said she absolutely loved that AI was coming because it stopped her doing the dull tasks and she could add value as an ambassador of the brand that she worked for. she was so passionate about her future in customer service as uh an adviser that would actually add value rather than just doing boring tasks all of the time. So that really is a a positive sign for the future because I think there will always be that human touch and the human adviser in in the value roles. But yeah, just replace everything else with the uh the taskbased stuff

doesn't need to be done by humans. It's it's like the industrial revolution, you know, with machines coming along to to replace the old sort of manual things. >> And that was pretty beneficial. I would say it worked out pretty well for everybody. It worked out pretty well for everybody. >> Yeah, we got food that we could afford to eat because they had a tractor rather than 10 people on a horse. >> Yeah, it's uh the old wheel. The um the wheel is a good indicator of it. It's I mean it's leverage. I talk about uh that there's a phrase that Steve Jobs called or not called he stated I think it was in the 1980s I want to say how he wanted to make the computer a bicycle for the mind and while I do believe that it's accurate that it has been I think

the AI agents are essentially the the bicycle for the they're the sports car for the mind at this point right I mean It's and then at some point we'll reach the the analogy to where airplanes and then rocket ships. We'll see. We're getting there. We're getting to the Yeah, that's the next natural step. Uh where do you uh find most just on a personal note, where do you find like most joy in like what you're you're currently doing, right? So it's obviously a pretty you said it's been an exciting like the most exciting two and a half years of of work experience for you. >> Why? How? As I said, I joined for the platform and stayed for the people and really engaged. It's the team that I'm working with are amazing. It's the best team in 30 years I've ever about 250. So department's

just over 20. So yeah, my my direct team, they're young, they're enthusiastic, they are so passionate about what they're doing, but they're also really curious. They're they want to know how stuff's happening. They're always researching the market in their spare time. They are the experts. I mean, one of the guys is a web developer, but he spends his whole time finding AI tools and what they're doing, and he's in all of these forums. They're just curious, happy, motivated people that just love what they're doing. So, yeah, that's what's really made it the tremendous amount of fun. And then working with, you know, founders like Phil and Sasha, they're so inspirational and, you know, they are remarkable people that can do a multitude of jobs being, you know, they started the company, just the two of them. So, they can code, they can sell, they can

pretty much do everything. We always joke that that Phil wears the chief hat for pretty much every role in the company. So yeah, it is the people we work with, but also uh what I'm proud of is the the recognition of what we've achieved in the market. So all of the analysts from Gartner to Forester to IDC put us right up there as a leader in in every survey that's ever happened. And then more importantly, all of the peer reviews from customers absolutely, you know, global win. They did the um critical capabilities with Gartner on the last MQ. uh we were the only company to be number one in every category and I don't know if that's ever happened in Gartner before but so yeah the proudness of of what we achieved with customers but then the the fact that customers love to be with

us and speak on our behalf so every year we run a customer conference in Germany we call it Nexus we had 40 people wanting to stand up on stage and tell their stories 40 separate customers uh for a company of our size that's incredible we ended up with about 30 and we even had to put people like Luans on a panel uh because we just didn't have space on the agenda But everybody wants to tell their story. Um when we've gone through investment rounds when the investors said who can we talk to we just given them the customer list and said you could pick anyone and the final bit of it is I'm probably jinxing the world now. We've never had a failed deployment. So we've never failed never. No for an enterprise software company that's amazing. So yeah I've probably jinxed it now but

that'sations. Yeah that was a good jinx. That was good. you know, some you got to do it at some point, you know, and then you'll get over the getting over the hump, you know, at some point it's No. Um I hope it I don't wish that on you. Uh but no, that's okay. Well, I I think there's a lot of people um I'm sure I mean I'm I'm on the younger side. I'm 27. I'm sure a lot of people my age, it seems like when they understand this market and understand what it means, they get really passionate because they understand that being on the right side of this is probably a good idea, right, with the with AI. >> Yeah. Yeah. My my son's busy my son who's 28 is busy building an AI version of himself to actually do his job just for fun.

These are >> well that's he's a business question. He's not a a technical person at all but yeah you definitely want to be on the right side of it. We've done quite a bit outside of you know obviously the the agent program that we were doing. Uh we quite often uh have school children coming to the office. We had a big day for for girls in it to try and encourage them to come and get excited about being developers and doing all the roles. We've got a fantastic peer group here of of you know all ages women that are doing a fantastic job you know in leadership and uh you know on the coldface on the coding and that side of stuff. So yeah they're all very inspirational. So yeah definitely get in on this side of the fence. >> Have you ever seen uh

the show Silicon Valley? It was a it was a it was a show about uh a tech startup in Silicon Valley. It ran through I think 2013 to like 2019 or 2018 something like that. And uh yeah, there was a consistent reference to the the lack of uh female coders and and whatnot. It's a it's a good show. I I highly recommend it. Um okay. Well, that's that's I think that's awesome. And just kind of as we're getting to the last couple minutes of this episode, is there anything specifically about what you guys are doing that you'd like to plug? Is there anything that you know you feel like is is up and coming that you're excited to to chat about or do you just want to um or any any final thoughts before we close this up? I think the final thoughts is that

the differentiator is going to be customer experience. So really focus on that. Uh and make sure that you're delivering exceptional service. So every single interaction that your customers have with you is a happy one. You know, regardless of whether it's inbound, outbound, whatever. Just, you know, let's make the world a better place by getting rid of uh customer service that sucks. >> You know what's funny? The there's a recurring joke in Silicon Valley. And this is back when I mean this is referencing more SAS stuff than AI agent companies, which to be fair is kind of an interchangeable uh term now because SAS companies better have AI in them. The um whole show starts with them at a party where they're at a uh funding party or like an acquisition party, I forget, and they're talking about how this company is making the world a

better place via and then X and then at the end of the and then they go to a conference like it's called TechCrunch Disrupt. that's in San Francisco and everyone's pitching, right? And everyone's saying we're making the world a better place with X. And I just love because it is kind of the universal like call of every every tech company. But I actually believe it like it's it was kind it's kind of a funny bit, but I do think you know if this is something that's really saving people effective man-hour to this level and it's who wants to sit on these calls like you know that not be understood and just kept so mad because it can't be good for people's blood pressure. No. Yeah. I mean, you know, the worst thing is when you're It It happens in in so many different directions. I

mean, think about it even from a brickandmortar store. Like I I I go to this local place by me for just like getting a tire change or an oil change or whatever it is, and their front desk has this weird thing where there's about two front desk people. Sometimes they're there, sometimes they're on a break. So then the mechanic has to go to the front desk and then sometimes there's a call that happens and you're standing there and you're like there's like way too many people at this desk and way too little people in line for me to not have been received when I walked up and I get that there's a phone call but like really and that little micro interaction can piss someone off to the point not wanting to go back. >> Yeah. That's you have a bad experience you know you

only get one chance. Yeah. Yeah, absolutely. All right. Well, we really appreciate you being on the show today. Uh remember guys, go to cognagy.com. That's cogni gy.com. Thank you so much for listening to this episode of the AI Agents Podcast and we'll see you in the next one. Bye. [Music]