Hello, everybody. Warm welcome to the Partner Marketing Podcast. In today's episode, I will be speaking with Glenn Calvert. Glenn is the founder and CEO of Kaizen, an AI client services platform for client services companies. Glenn and I will speak about how AI can help client services teams. We will, of course, speak about AI in general. And because that's the Partner Marketing Podcast, we will speak as well about how AI can especially help in partner marketing. Very warm welcome, Glenn. It's great having you today. Matthias, thank you very much for having me on. Looking forward to this. Great. Maybe as a first step, you want to introduce yourself a little bit for our listeners? Absolutely. Well, Matthias, thank you very much for having me on. I'm looking forward to talking to you.

So I'm currently co-founder and CEO of Kaizen. We're an AI platform specifically for software. We're one of the service-centric organizations. Our vision and our mission is to build the ultimate AI assistant and AI co-worker for anyone in enterprise relationship management, so client service teams, account management teams. We've got this vision where I think in the past, client services and account management has been very much a very manual, very reactive profession. And we think it should be much more automated, a lot more strategic, add as much value to clients. And we think it should be much more automated, a lot more strategic, add as much value to clients. So that's what we're building. I started my career, though, actually at Tradedoubler. So I finished university, went traveling. When I was traveling.

It's always a good start. It's always the best start. All the best entrepreneurs started there. So yeah, finished traveling, came back and thought, what am I going to do? And thought I need to start understanding digital marketing and got a job at Trade Doubler. And that set me off in my career. The very first startup, straight after I started. After that, part of the founding team was struck, one of the first personalized retargeting companies. Back then, it was sort of crazy to see sort of personalized adverts on your computer. And we sort of helped power that. And then I built and was found a CEO of Effective, which was a big programmatic services business that worked with the agencies and the brands to help them do display, video, programmatic marketing.

And then took some time out. And now we're here building. Building Kaizen. Now, Kaizen, finally, you are in AI. It comes with the promises of increasing client engagement, increasing revenue, increasing productivity. We work with Kaizen. Maybe that's interesting for everybody who listens. Our client services teams are using that. Now I get, of course, how you can use AI to increase productivity, like creating reports. For example, in our case, suggesting partners for different affiliate programs, maybe do some kind of gap analysis that helps. But especially around increasing client engagement, increasing revenue. Can you describe, and of course, on the productivity side as well, can you describe a little bit on how you do that? What is the tool about? How can it help client services teams? Absolutely.

So, yeah, I'll give a bit of context as to sort of how it works. First of all, sort of any user just activates their AI assistant by logging in and granting access to calendar, email, and the chat systems they're using. And then from that moment, the AI is basically helping a variety of automations, writing your emails for you, writing your meeting notes for you. But in actually plugging in and seeing the client conversations, the client communication, it can then start to harness all the collective intelligence within the organization. What's trending across all of the conversations? That we're having what are we seeing across these different clients? And then from there make recommendations as to, uh, as to what to do. So we try and tackle um, two areas.

So two core areas for any sort of client service team. One is the automation. I think we'll look back, you know, in a decade from now, we'll look back on the idea of updating CRMs and doing scheduling, um, and responding to the draft emails, you know, we'll look back on that and think, 'Oh my god, that was crazy!' We spent so much of our working lives just scheduling a meeting or updating a CRM or project management system. That will absolutely be taken care of by your own AI. But I don't think that all those automations end up ending into a world where we're just all less busy. I think we'll still be all really busy.

Just the nature of the work changes and that leads us to what we really want to focus on is how do you take all this collective intelligence of activity and empower client service teams? Anyone in N2 prides relationship management with more knowledge and more information to help them be better at their job um, so I'll Give you an example, so can we can we collect can we do deep research every morning on every single element of your client, their sector, their competitors, um their news and bring this into the into the inbox of the account manager, they don't have the time to go through every single one of their clients and do this research, so again AI can take care of that for them.

Um it could be in terms of the conversations they're having, say like you know ways they could improve the value they're adding or the questions they're asking um and just help them with sort of like commercial conversations so we take all of the all of the activity. That's happening with uh, with your clients. We take care of all the admin; we write the draft emails for you; we sort of update the CRM for you; take the meeting notes for you. Um, but I think they really enjoy is those sorts of insights that come through and help them be better at uh, at client service. I think in the future, what will it also enable is you being able to sort of create um duplicates of each account manager, so all of those all those um

issues that clients have got or value they want to receive but they've sort of you know, you know, got a one account manager who might be constrained with time um because they've got multiple clients to look after so how can you add more values of clients 24/ 7. I think that's where we'll start to see these AI co-workers develop, and every account manager will have their own agents working for them across all of their clients so you can provide 24/ 7 value to every single one of your clients. Can you maybe describe a little bit on how you do that right? So we, for example, have a larger enterprise client working internationally; we have people in all the different markets being in touch with the client has different contact persons on the other side, yeah, there's maybe a little bit different strategies per country, there is for sure a lot of conversations ongoing, hundreds of emails probably per month going back and forth.

How, how exactly can you describe a little bit how your tool helps to manage that more efficiently? Yes, so technically, how it works is you basically are looking at every type of communication that happens with an individual; so you would either have it on a call, you'll understand their email address as a participant on an email, you've got their email address and what you can then start to do is build a model specifically for all of the communication happening with a particular client, and then also all of the Individual stakeholders, so then the the the model you have assesses what's happening with that particular client: what are their high sentiment moments? What are their low sentiment moments? What's the products and services they're buying? What's happening in their world?

What their objectives are? So you build a specific model for every single client, and then from there you can understand really what's uh what their needs are. Each individual user also has a model that's looking at how they communicate, how they read, how they write; so we start to draft their emails for them, it's specifically based on their tone. And their style with knowledge on what the client cares about, so it's understanding from the last sort of six weeks of conversation in this particular market, what does this client care about, what their objectives? So we take all of this communication data, have different models do a variety of different things. Um, now we have a framework that works pretty well, so over sort of two years being closed beta, we wanted to say to account managers, look, we want to take you from a reactive, stressful type of job into a sort of proactive, strategic um department um, and they said, well, look, there are a variety of inputs that We know lead to good account management, and we know lead to happy clients, which leads to more revenue. So we've created this framework, it's called the CARE Framework. The C stands for Client Knowledge-so how much knowledge have you got on the client, their sector, their stakeholders, their competitors, everything down to like the birthday, down to these most the macro movements that particular client? So the C is for Client Knowledge, and what's the depth of that? And again, you can start to model that and understand that by looking at the communication. The A is the Activity-so it's uh it's no it's no surprise if you speak to people Regularly, if you respond fast, if more you know, engagement with decision makers you end up normally having a good relationship with the client; The 'R' stands for the Relationship and the Sentiment Score.

So, we're looking at here that more behavior um does your behavior type match up to the behavior type of the of the stakeholder? And the 'E' is Expansion and this is normally the sort of left last. Are we looking to proactively grow and scale our clients' business? And you can do that by having a good knowledge of what's happening and what's important to them, but also asking the right qualifying questions so the AR will start to Spot things and say, 'Oh Matthias, I can spot you! You've not asked these questions that could be relevant for that particular client and lead to expansion. So the care framework sort of is embedded in everything that we do and it's basically looking at this in real-time across every single client relationship and making suggestions through to every account manager saying, 'Here's how you can level up the the client health score or the care score with that you manage, you measure um activity obviously you look between the lines, you look at the tone, you look about the topics and the topics that have been discussed. You make some Kind of gap analysis, and I come to some kind of conclusions, um, where are we with this client? What might we have missed? What could we do next? What could be the next email I draft? You something I make you suggestions. Maybe we can take a step back on how you do that so you're integrated in a client's business.

So our business for example, um, and for this particular relationship for a specific client. Basically, gather the data of all interactions is that right? So you get the emails, um, you transcript any kind of calls or meetings, yes, um, you get access to CRM data, correct, correct, and also, um, if there's any sort of Slack messages or Teams' messages, they're also incorporated because the stakeholder is normally the anchor is like an email address you can use the same email address center so automatically collect the same email for a particular client then you can even connect to all that sort of communication data and put it into one folder so users don't have to start to manually move things around or update things they just go about their day talking videos emails etc etc and Kaizen's understanding all that communication data bringing it all together into into one place and then if a client wants to have their CRM updated in real time you can.

Take that communication data and you can push information back into the CRM, so you get this sort of push-pull relationship; um, that's that's going on but it gets really, it's it's more; it gets more interesting when you look at the collective activity, so you start to look at, okay, what's happening with these sort of three clients in this travel sector and we seem to be providing a good value to them and good service them and they're spending more money with us, how's what's happening there in terms of the activity, the account management versus these clients over here and you can start to then benchmark what's going On across the, uh, entire, the entire of the business, and find and find opportunities.

And we've seen this with multiple clients, you work with, in particular marketing agencies where they realize how much time they spend in conversation around strategy and ideation versus just reporting. Naturally leads to more opportunities and more client growth because you're sort of being seen as a thought leader to grow their business versus just falling back into a reactive 'here's reports' and 'here's the activity we've done for you', so you can start to then optimize your go-to-market teams, and your, and Your uh your client-facing teams to provide ultimately more revenue for the for the team, there is some obvious things right that I can imagine right away, like um, for example, if we work on a set of clients, let's see, um, we work on a set of 100 clients with fewer technology you can measure stuff like somebody has not talked to a particular client for some time, right?

I guess that would ring a red alert already. And from there, I think you can go all the way up to being really sophisticated um in what kind of findings you have, like for example, suggesting us a a a new tool upsell something like that where are you there how would you say Um, where are you there because Kaizen is a startup, yeah, right? So, um, you how long are you around? Uh, so we had two years in closed beta and we came out of beta last July. So, we're coming up to one year out of out of closed beta, so we've sort of been around three years, three years old. It takes quite a while to build technology like this because you've got to build everything specifically with security and privacy in mind, because you are, you know, you're processing and housing a lot of, you know, very, you know, critical business-specific data, and therefore, you've got to be very um, secure in your approach to uh, privacy and access.

Works within the system, so yeah, two years of understanding what the real the features need to be and the value is, and how we're going to prove that to the clients, and then building a way that is sort of privacy and security first, ready for um, for scaling up. But you're right, you know there are there are things that are easy. Well, you know, you've not spoken to decision maker for a while, there's a red flag. Um, it's already it's already a a something great, right? It sounds super easy. Um, but it's already a big achievement to be able to tell something like that, right? I from my own experience, I know that it really helps. Thanks, man. So, the uh, where we want to get to with your question like where can it go?

I think that we're already providing value in little things you wouldn't have expected. So we thought that by doing all the automations of writing your notes, tracking your action items and your emails, there's a lot of value there which there is. But then what we didn't realize was actually the alerts that come through each week to account managers around their clients and what's happening in their world and their sector and their news actually as much value there because they're just so busy, therefore if you can help them okay before they've Got a conversation with a client, we're doing some account planning, and if you can give them a bit of information and insight, and

obviously because all the breakthroughs recently with research deep research models really help in this regard, so I think we'll go from being what feels like quite static at the minute; we'll send alerts inside the Kaiser application but we'll also do it inside where they live, you know-the inbox is really the HQ of an account manager, most of it sort of-most of their time is spent in the in the inbox and uh so we bring alerts there. We want to get to um and hopefully this year. is the their assistant is a fully interactive co-worker so you're to speak to it communicate with it um it'll be able to respond ai agent it'll be a full ai agent that you can actually converse with whether you want to have a conversation with it in the app because you're at home or you're during a conversation and you want to speak to it as well so you're going to do it go and get um get information for you or do tasks for you so i think by the end of this year it will be a fully fledged sort of ai co-worker that's within the team um and you can you can converse with and it will be very and it'll be connected to critically connected To all of your internal systems, so that it can go and bring information in for the account managers not to waste their own time on that um, and also populating systems are that are slowing them down.

So if we were to get the AI co-worker into a great place, then every account manager is then truly strategic; they're really thinking deeply about the clients the client's business, and how to provide more value to them. If they do that, more revenue comes through versus having to just go and you know search through internal systems, populate things, spend their time doing admin like responding to basic emails, and are they, you know? are they free for a meeting at three o'clock all that stuff the ai should take care of when you think that to the end there is basically no limit right so you could have your ai agent even sending emails to the client directly at certain times you can train that doing on standard tasks like for example sending a report but you could have client qualitative messages at certain intervals as well I guess your technology for example could measure what would be the best logic for sending sending out proposals yes another proposal you wait three days then you wait another three days you remind I guess you could measure very Good, when would be the best point for the first reminder? How would you do that how would you phrase that? And when you think that to the end, there is basically um no limit. So, do you think AI can take over all tasks in client services or client management at a certain point? Very very good question! My short answer is no because I think that what happens is that what's currently account management clients will just evolve um so here's our prediction is this and I think this is sort of well grounded because it's been requested by our by our clients so you I think

will happen is that there will be a bunch of services that you know ai is going to is looking at the whole services industry and saying how can i provide value to to to uh consumers here in a way that sort of may have been blocked to them in the past if you wanted sort of really good management consultancy you'd always go and spend a lot of money going to one of the big four to get that and now actually with ai models you can get a lot of management um uh you can get a lot of consultancy by just using the models so i think what will happen is services will evolve account management within that will evolve and you'll end up with the expensive high value account management that customers are paying For but some customers who maybe don't have big budgets will still want your services and your products, um, and then, but maybe not get the full suite of having an account manager, human, coordinating and managing it for them, and I can't imagine it's in the future because our clients have asked us this-it's said we could spin up what effectively looks like a a chat bot, you know, it's our own company's account manager to take care of the lower-value clients, the clients haven't got the big budgets but we still want to provide some services to them, and then, the larger clients who want a full suite of account management will get the.

the the ai enabled ai enhanced humans doing that so i just think services will evolve i don't think it'll end up with a place where it's fully replaced i think certain types of products and services will be delivered through a company's ai um and and remove that from the humans because it would be too there's no there's no uh there's no value in them looking after that that workload um and then the actual the the larger clients will be taken still be managed by the by the by the humans and i think it will also evolve if you look at any sort of tech you know look at any um tech platform breakthrough over the last sort of 30 years It doesn't, it doesn't you know it doesn't replace jobs, they just change you know.

If you look about, look at social media now and the creator economy so basically what you were doing if you were working in magazines and magazine advertising 20 years ago, guess what you're probably not going jumping into that industry right now, you're more going into social media management and uh and those types of roles. So the nature of the of the work changes I think exactly the same will happen because of because of AI. It won't fully replace roles just the nature of the roles will change but it could change fundamentally right because when you when you think it to the end um basically all tasks in client services except of the personal relation of course um could be potentially replaced by a ai tool i guess well i think um if i think about kaizen and any any sort of services that we buy and use i still want to have the conversation with the individual i want to know who they are what their company stands for i want to have trust with them i want to know who they are what their company stands for what their passions are and what their ambitions are and then does it fully align with with us i don't think we'll end but it'll be removed down to a just a pure automation i still think the human element this is actually really fascinating I think the if you wish to double down on building an ai that enhances account managers and makes them very very good at the softer skills um that's where the real value will will come because you're still going to have b2b relationship management just certain elements of the business that you're going to have to be able to do that that relationship will be automated um or certain or certain areas will be fully fully automated but i i think that um when you're going to have when you're going to have partners and vendors you want to work with you want to Look, the person in the eye, meet up with them, trust them, and then they can use as much AI within their organization that's offering as possible to create more value for you and keep the cost low.

But I think the human human element will always remain. I think still the role of a human account manager will change fundamentally. Um, which I guess brings us to the question of accepting that... What is your experience there from the conversations on the one hand with these account managers? You're working with organizations, you're teaching them to use your tool, you're teaching them to to use AI obviously, with the focus um on productivity increasing. Efficiency, um, taking tasks into AI which obviously goes away from the account manager and then on the other hand, you have um their clients so the brands potentially yes, um, I say potentially could be other kinds of clients as well, yeah, that they are working with, that might get automated emails, automated reports, automated replies.

What is the acceptance level on account manager side, so user side from your perspective and then target side, so brand side, the receiving end where are we there, yeah, very good question, so I think what I what have what we've realized and what we've seen, the the most sort of you know, visual in your face elements of any of the sort of the ai assist that we've we've created is definitely on the video calls so that's the assistant that dials in and joins you on the on the calls um and each company has their own assistant that's sort of working for the for the account manager now it's two and a half years ago when we started to really like get under the skin of what this technology would uh would enable and so what the user experience was recording was quite it wasn't common it was still you know it would be unusual for an assistant to be dialing in now i think that's completely different i think especially in the us in The US, it's sort of it's the other way around. It's a bit odd if there isn't a call recorder on the on the conversation um, whereas in Europe I think different markets are still on a different trajectory of uh, of adoption. I said I think call recording is over just is naturally seems just more and more common now and again fast forward five ten years every company will have uh, a call recorder on the iOS and AI agents doing a variety of tasks for them downstream from the conversations so again, everything will become recorded just to enable the automations downstream.

So I think the call recording, I think has been. the one area we've seen just slow and you know adoption just increase um is that mainly because of security concerns or data concerns um or some kind of awkward feeling just i think it was more i think it's just more the awkward feelings certain individuals would be like i've got a cool recorder on this you know what you know it it's certain individuals are totally fine with it and somebody oh this is you know a bit a bit alien and a bit unique um so you have to make sure that all of the user experience and the features around it enable people to have

different types of you know which calls it joins automatically which one it doesn't join how to remove it, um, so then we've seen that adoption level slowly increase and I think the um in terms of the education to the users once they start using it, they don't go back, you know, we have very, very good um retention because once you start to have an assistant that's helping you with writing your emails and writing your notes, it's very hard to go to go back to the way it was before. You oh my god, that's uh, that's um, I'm not gonna, I'm not going to do that, so we see very good adoption there now. The client side what's been fascinating has been and I didn't, we didn't expect this; we imagined the first iteration would Be more like a client portal where it would be a place for the client to log in and be able to speak to the AI as well as to get reports and to basically understand you know questions and answers from from prior conversations, um customers who are in for us, a lot of them in the marketing service space, therefore the brands have been positive to companies using Kaizen, because they're basically the the company normally has some level of service level agreement in terms of we will respond fast or we will you know we will be on top of your business, and therefore they turn around to the we're using a client service platform. To ensure that we are on top of your business, our team are fast, efficient, got access to information.

We can stay on top of the SLAs for example if you know we've got a three-hour response time SLA; we can see that within the in the platform so um it's been put in a positive light in front of the brand to say, 'Oh, look! Trust us, we're going to be very much in terms of giving value to you and staying on top of your your business to the extent actually putting Kaizen into their new business pictures. Saying our client service team using AI to make them really quick and really fast, which obviously which is which is uh which is great for us, which. Reminds me a little bit of when we first started using ticketing systems, yes neither the account managers like them working with a ticketing system nor the client likes to have a message from a ticketing system.

Now, we just re-evaluated that again in a there was a completely different conversation. But the account managers like it very much because they can work in a structured way; some said as well it protects me at a certain level. Clients like it as well because they have to track and they know they'll get a response right, so that has changed. It was actually these discussions we had for quite some time, right yes. Yeah, I know exactly quicker now with AI, I think yes you think so yeah, it's much quicker and obviously you can do a lot of the the responses are not only you can do a lot of automated responses but not only save the account manager's time but the responses are better because you basically are we used to say rely on the account manager's uh knowledge and experience and someone who's sort of one year into the job be different someone who's five years into the job, but now when you're suggesting answers and suggesting what to send back again it's based on the collective intelligence of what's the best have we seen this question. before what's the best answer to it we've seen collectively in our business we should suggest that and put that back in front of the client so i think it's a it's not just the speed and the efficiency that's great for the client and great for the account manager but you actually improve the quality of the answers and what's and the information being shared by harnessing all the people you know run through your Fl excuses during an interesting process i think and you and everyone has that an example that they've had mentioned in the discourse but it's a different będzie pitch the way data is meaningful in this type of leadership There's a lot of information about how to do data, and hello, all of the data is useful.

It's seven of the virtual harmonica which can explain why the score was higher. And you can use that, and if you need to take on different FaceTime calls and do various housekeeping; the main challenges for the companies using your tool not not while using your tool but by applying it to their own business getting the most out of it, yeah, it's a very, it's a very interesting question I think. To our experience has been that we're very early on the adoption curve of AI being implemented across companies so if you were to walk into a business with A project management tool or a CRM, it'd be very clear what that product is and what it's used for.

It's a very well known category now when it comes to AI assistance and AI agents and AI platforms, and how companies think about what AI means for their business moving forward-what's going to be fully automated, what's going to be augmented, what new products and services can we create off of the back of this? What's going to fundamentally change how we operate? This is all very new to companies, so they're really trying to understand: what is the AI agent infrastructure that we're going to need? How are we have to think about That um what do we want to do with it, what do we not want to do with it? So companies still just on the journey of, like, from a board level thinking, how are we going to deploy AI to improve our products and services, our delivery, and our operations make more, make more revenue, makes more productive, and they're all just I'm trying to understand that so when Kaizen comes along and says, 'I've got an AI assistant for your client service team you're basically going to help them be quicker, faster, and make your clients more engaged and spend more and spend more money with you. That ticks a box; like, that's a pain point you're going. To solve it, but what box do you fit in? Is it CRM, is it a sort of AI platform? Are we going to be, is it test and learn budget um is it you know AI assistance budget which is still being thought about so that's the a real challenge.

So where how do I think about this being pushed into our business and what category if you know what line item in a budget sheet does it uh does it um does it come from um? And then you, you've got to your to your point which I think is an important one is how does this fit into our current way of working so we can take all of this knowledge all of this this data, how does this feed into our BI and our sort of our reporting? How does this feed into our CRM? How does feed into our time tracking, um, and that's an opportunity for us because there's a lot of every company's differences to what their focus points are.

And then we can then sort of adjust, sort of the outputs and the reporting for them based on that. So I think the company is just still going to figure out what where is the what are we trying to do with AI? Therefore, when we're going to start to use tools and services and build things ourselves, which budget lines that come from when a company decides to work with you? What's the entry hurdle so you collect all this data, emails? Meetings, conversations, etc., um, how do you integrate or how do I integrate with you? Yeah, it's um, it's we, we very purposely made it as easy as possible. All it takes is an account manager to sign in and then and grant access to their calendar, their email.

And then from that moment we look through the inbox, who they have most conversations which companies have most conversations with, that's probably the clients. And we create the client list from there. And they've activated their AI, so on the next call they have, they'll have their assistant or the next email that comes in, he's gonna write a draft for them. So it's um, it's very Easy to just to get going, um, and we made that important to make sure that the the end users, the account managers, the ones who are going to be the ones who are going to be the end users, we really want to empower. Find it super easy, and then from there, their boss might say, actually you know we'd like to train the AI specifically on our products and our services, these are the things we'd like to see in the reporting, and then we start to then from there build the platform around their needs.

But getting going, so we deliberately made it super quick and easy within one minute an account manager is up and running and getting and Seeing value, you get a lot of data and data security. I guess that's one of the key topics when engaging with any kind of customer from your perspective. Oh absolutely, I think it was absolutely table stakes for us when we built Kaizen to be built from uh the the highest levels that you could have from a privacy security standpoint. So when it comes to infosec, you know, speaking to an ISO compliance um the hard the the challenges is to do the the hardest part is to do with the permissions of different users so in your organization we have different permissions levels around different systems either in the CRM or with clients.

Access so we house all this information, it's all encrypted. Your own particular instance it's encrypted; no one at Kaizen can see any of the data, um, it's basically your own data. None of the data is used for any model training or model usage, um, so everyone works just knows my data is totally secure and private. And then on top of that, the users only have access to the right information that suits their permission levels. It's probably the hardest thing to get right. So, for example, you would want to jump into the chatbot and ask questions globally about any client; and you would as a CEO, you'd be able to say, 'Okay, well I want To understand that particular client in that particular market, what are their top three objections?

You can go into the chatbot and you can ask that. But you wouldn't want someone who's maybe just started the role in a particular market just to jump in at a global level and speak to the chatbot about a client and another another um another market um so you have to basically give data access and this is where it gets quite complex with the the the vector stores all this information of who can access what information, what can be retrieved a particular um based on the based on the question that's been asked. Again, it's just It takes its not it's not an impossible challenge, it's just when you're building software one of the hardest things you've got to build so why when you, when you, a lot of AI uh you're sort of in the news, it's around sort of chat GBT wrappers you know it's easy to get startups going now and build products on top of the APIs kind of but if you really want to build enterprise software you've got to consider all of these use cases um and these aren't easy to, to build so I think if you get um there are real modes of defensibility for companies building an AI by looking at the scaffolding you need around the different data you're going To use different models, you're going to use and security and permissions, uh, built within it, and that's where a lot of the the modes of defensibility is, and with what um AI technology are you working at Kaizen? So we, we're interoperable with all of the main models so we work with Open AI, um, perplexity, um, anthropic, obviously trying what we can do with with DeepSeek as well because there's a lot of interesting uh breakthroughs there as well as our own models.

So we, our own model in particular was built around sentiment classification, so understanding the sentiment and the high and low moments of of interactions with clients. So, obviously a lot of big part of what we do is understand what drives high sentiment moments, what drives low sentiment moments. We do a lot around sentiment analysis, um, and then we're interoperable with the different models for different things. So for example when it comes to client research and what's happening in the client's world, perplexity is very good at real web analysis and summarization. When it comes to content creation and summarizing action items, Open AI is very good at that. So we use different fine-tuned models for different parts of the work. Very good, thank you so much Glenn. Maybe taking it on another.

Leveling up about AI in general, right there's a lot of hype about using it. I use it myself almost daily; it helps me a great way, actually. Yeah, I'm still um always surprised but like uh, very positively right! It does do a good job... Um, then I had a podcast recording with Rob Barry for you about AI and partner market marketing in particular, and that was really interesting because I learned a lot about um the the absolute core – what is the essence? Which is the data. But about how data is stored, and usually data is already because of approaches before stored in very structured ways, which stands in the way of using this data.

to work then with ai so on the one hand there there is a lot of wipe there is new stuff coming out of uh every day and there is there is new headlines and and so on and so on um now you it's probably the first real conversation i have with somebody who has a real tool that you can actually use how would you see say where where are we there like in the in the great scheme of things you have all these uh um ai platforms and you have a lot of talks but how much is there in practice already now and how much is just like talking yeah very good question so i i think if you're building any AI product now having a working prototype is is table Stakes it's it's it's okay to talk about what the what the theory is, but the reality of building AI tools now you can very quickly get at least a working prototype um set up and then you can get your prototype going and show here's what we're trying to do from a data input data output here's what we want to automate here's what we want to enhance or here's what we want to enable. You get the prototype going pretty quickly, um, then you've got to understand the use case very specifically, so it's your point there around the data, what's useful and what you know you had structured data and what AI is very good at unstructured data i think with with kaizen we had to understand what are the sort of questions that people are going to ask and what are the information they want to um receive and very quickly it became apparent to us that actually the structured data was still very very important and then we had to understand that what are the important so a lot of the questions would relate to things like what are what's the this decision maker what are their biggest objections to our to our business so in asking that question you need to know who the decision maker is that's now structured data you know the individual you understand what their role Is it about to retrieve all the information specific to their calls and their emails that happen with the with the company now, then they do objections that's not on the qualitative data and using the answer what were the

objections and were they real objections were they complaints about the weather were they complaints about business um and then recency so again you don't want to have the information pulled in from two years ago you need it from the last the last 30 days so that when we actually build the when we've had your approach building Kaizen it's a combination of the the quantitative structured data. Of clients, stakeholders, time frames, date ranges, and then interspersed with the qualitative as all of the communication data and the information about their... about their world. So you start with the use case: you say, 'Okay, well what is what is the use case for us?' There's a variety of automated output there's a bunch of things that come in from the questions and the chatbot usage.

Okay, what are the types of questions going to be asked and how do you structure everything to then give you the best output? And that's then that's the fine tuning. You start saying, 'Right, okay, I know we're going to get these questions more often than not let's fine tune the system and say here are good examples and bad examples for those type of questions and output you build the you can build the ai output based on based on that and where are we like when i want to go out in the market um how many tools are there that are really in use and and really already have an impact on businesses yeah what would you say i would say so if i think about our experience with the clients that we're working with it's the the variety of um adoption is is great so we work with people who are already building proprietary internal workflows very specific to their business so for example some agencies have built um media planning tools and research tools based on all of the media plans and research they've done for each sector in the past, so when where their team are using doing a media plan it's going into the history of sort of you know the last 10 years it can reference that and come up with ideas and and uh and help the team um so I think it's like truly proprietary um unique technology that they can also use when it comes to new business pitches and show how innovative they are, and then you've got some who are still on the adoption curve of maybe just using you know just having individual conversations with ChatGPT um Or, anthropically to basically help with with brainstorming and ideation so the the you've got some companies who who are still unsure, what does it mean?

How are we gonna use this? And you've got some companies that are very much like, 'we are AI-first' and we're building workflows that are specific for us, but also thinking about products and services that are put in front of clients, they can monetize themselves. Maybe it could be like image creation and content creation and campaign creation. There's very much like, this is like an AI-first product. It's a new product or service that they can offer. But I think for the most part, everyone's all of our clients; they're deep in with the standard models. They're using obviously Kaizen and very business-specific tools. If they've got an engineering team, they're probably using some of the code creation and like GitHub Copilot and Cursor. If there's anything in the client-facing team, it's probably a CRM with a Kaizen built on top.

And then you've got the teams with innovative heads of IT, innovative CTOs thinking, well, actually, how do we start to build something quite interesting proprietary using our own data that solves a real use case, but doesn't suck up loads of our time and energy and is actually gonna be core to what we do. So on the one hand, you could say still early stages. On the other hand, so clearly visible way to go. Yes. But it will be inevitable. Yes, it will be. I think it really comes down to the leadership team. So where are they? Where's the leadership team from an adoption point of view and a comfort level with how they want to innovate and push in particular areas. And I think we've got-We've got certain ways of working that are established now.

Exactly. There are some marketing service agencies we work with who have got particularly innovative, sort of like co-founders, who are looking to constantly push the boundaries and take very unique products and services and ways of working into their clients. And they're leaning into it very hard. And then you've got some larger companies that maybe, you know, they haven't got, there's a lot of legacy and there's a lot of change management required and it just takes them time to think things through. And again, the larger the company, the more security and privacy considerations you've got to have as well. Great. You have a background yourself in digital marketing and a bit of a background as well in partner marketing. How you see the use cases for partner marketing in particular? Yeah.

Yeah, I think it's fascinating because when I first started at Trade Doubler, it was first to look through what we were doing for clients. It was a lot of search affiliates and a lot of destination comparison affiliates. And that was because that's where the internet was and consumption was at that point. And obviously now we're living in a world where it's the creator economy. It's about video. It's about everyone being their own media owner, having their own YouTube channel, having their own following. And I think that only going to increase, especially on the B2B side, is going to give you that kind of link with what LinkedIn are doing and their LinkedIn focus on video. So I think the nature of partnership marketing then evolves with what the consumption looks like.

And I don't see that slowing down in terms of the creator economy and what that's going to enable. I also think there's going to be a lot when it comes to how AI will be used is how you find the right partners and the timing of who to work with and the research and the the identification of the right partners to be working with is going to improve in terms of accuracy because of AI. I think there's a thing, I'll be keen to get your thoughts, Matthias, on this, that the internet has been consumed by consumption. And I think we're moving to AI agents being prevalent across the web, where you can say to your assistant, 'Oh, can you go and please book these flights for me?' And the assistant can go and crawl the web.

It can research. It can come back with the answers. And then so websites will start to be created with agents in mind. And people won't go and do the work themselves, go to a comparison site or do the searching. They might just use an assistant or even agents to go off and do the task for them. I think there's a huge question that enables them. I think we've gone a little bit further up the channel, right? So partner marketing in the past has always been very last click. There was a lot of business models just before the conversion. I think it will move up, right? So I think down there, there might be a lot of AI impact on how to automate that, having agents just like asking for the best offer or something like that.

But when it comes about generating interest, as you mentioned, the creator industry, there is a lot to do. And I think that the whole funnel. Yes. Marketing funnel will maybe change a little bit and will shift. Yes. That's a bit like kind of like how I imagine. Yeah, I agree with you. There's a lot of theory here. It's a lot to be proven out. But I think if it does go down the route where I don't go and go on Nike's website and Adidas website to look at the, to say I want to buy, but if I was actually to say to my perplexity agent, oh, can you please go and find me the, you know, the best trainers in this price point and come back to me with like five options.

All the same when it comes to buying any, any item, that's a very, that's a huge change in how we will purchase. Now that's a lot of theory as to would I, would I entrust that to agents to go and collect the information, bring it to me, but I then go and review and then, and then maybe complete the purchase. Um, but I think if that does happen, that means there's a, that changes a lot for the mid-market of e-commerce and, and brands totally because you, you, you, you become dependent. Dependent on whether my products been found by agents, have they been brought into the inbox of the, of, of the user, however, brand creation, you know, brand creation recognition, making me resonate with them will come through, um, the, the, the work that's top of funnel and, and, and with, with creators.

There is a lot to do. This is, um, there's a lot to do. Uh, what are your main topics are now for the, for the coming 12 months, I think you need to make some tough decisions there. Yes. Yeah. So, so what we, we, so with, with, with Kaizen in, in, in particular, we are, um, when it comes to, we're at a stage now around amplification. So how do we take the brand to market the company to market? How do we tell the story of, of what's possible to account managers and client service managers, anyone in enterprise relationship management? How do you tell the story of productivity gains, but also improving, you know, the, the knowledge you've got on your clients now to, to, to communicate and engage with them, and ultimately the revenue gains.

So we've got to, we've got to now take this to market and I think we'll do what most companies are doing. It's like, how do you go beyond, um, uh, last click and how'd you start to think about creators and content and do it in a smart, innovative way, but critically have a viewpoint, like what's our viewpoint? What are we known for? What, what's unique about us? That's our job is now spend the next 12 months, maybe just amplifying that, uh, that narrative for a product point of view. The AI has got to be fully interactive. It's going to have full access to all of your systems. It's going to be a the most uh important coworker to your account management team.

Um so we can just totally revolutionize the way that um value can be delivered to clients and the speed at which it can be delivered. That's our that's our plan for the next uh next 12 months. Super interesting. Now we're very happy to be on this route there together with you right It's super exciting at the end We're coming to the end now Yes If that's if that's okay Glenn at the end I asked my guests always the same three Okay Final questions. Um, if you don't mind, go for it far away, far away. Good. All right. I would want to know, it's a very much, it's a bit selfish because I, I always take that for myself too. So what would you say is the best book you ever read?

I need, I need some recommendations. Okay. Okay. So very hard to come up with just a single book. I can, if I can do it by category, then I'll, I'll do it by, so if I, if I was to, for a business perspective. Um, the hard thing about hard things by Ben Horowitz is, is, is, is right up there for me. I think anyone who's involved in building startups and companies, it's just really, really hard and really, really difficult. Uh, and that book is, uh, basically just is, is a good reminder of how difficult it would be. And everyone's experienced the same thing. Um, connected to that, I think would be more of a history book.

Um, so man's search for meaning by Franco, which is a story of a, of a, of a guy who was in a concentration camp and how he dealt with that. Yeah. So it's a, it's a deep, it's a deep book, but there's a lot that you can, you can learn from it and, um, uh, and, and take from it and, um, something that'd be a, a bit lighter. I would say my favorite fiction book would be, uh, 'The Old Man in the Sea' by Ernest Hemingway, which is a very simple story of a guy fighting a fish. Yeah. But that's lovely. That's, um, that's an impressive rate. Yeah. I know that one. I will take the other two. I only struggle. Okay.

Which is like we do the podcast recording biweekly and I can't catch up with the reading, right? So I always get like really, really good tips. Um, the other one is a bit easier, um, than for me to adapt to a tool or an app that you would say you can't live with. So I would, well, I think the founder of Trade Doubler went on to fat found another very interesting business called Spotify. And I think that Spotify, for me, if you were to save all the apps I use the most, I think, user takeaway would have the biggest effect on me. I listen to a lot of music while I'm working or outside of work, Spotify, that's that to me is the number one app and I've managed to maintain I think a good user experience despite becoming quite big and uh, and a large business, I'd say Spotify.

But then from an AI perspective, it's become totally natural for me now not to open Google but to open up Perplexity or to open up Claude. In particular, I like Claude; I prefer Claude over OpenAI. I put around tropics news experiences but so yeah, Spotify, and then one of the um, and then one of the uh, the the AI apps, and we'll probably go with um, Claude. Very good, thank you. And the last one is a bit more on the personal side about me-if I wouldn't do all of that going into AI working with digital marketing before um being an entrepreneur, a founder, take all that away, yeah what would you be doing? Would you be the the young man on the sea fighting a fish? What would you do well?

If I had this, if I'm, if you grant me the permission to have whatever skill sets I wanted then if I if I had the choice it would be a professional golfer. Um, if it wasn't a professional golfer and I still could and I had the skills, it would be a lead guitarist in a rock band so I think Keith Richards, rolling. Stones that would be, that would be me. You play the guitar; I do play a nowhere near good enough to be on stage, but um, but occasionally so, yeah. If I couldn't be a golfer, I couldn't be a rock star. Um, and I had to do something that was not inside you know, AI, and uh, digital marketing. Um, I would probably choose...

I'm quite into sport, so probably something to do with the in, in relation to AI and coaching and development inside the world of sport. I think is quite interesting; that would be good. That's a way how to get close to the teams, yes, and you're talking about that too, yeah, yeah. Okay, good. But you play golf, right? I do, I do. Okay, yeah, which There were constant the constant journey to, can be a scratch golfer I think we'll get there eventually. I stick with tennis so, I nice I stay there no good if I can if I can throw the question back at you, then what would be the uh what books do you recommend for me um now you got me

I'm I'm not prepared enough for that for that at the moment I'm sorry for that too early too early on a Monday neck too early on a Monday yes and I um I can't keep my mind now I'm in podcast recording and then AI I can't switch but next time we see us uh when I'm over in London um I will have a book recommendation for you, thank you for sure, thank you man this was great. Speaking to you, thank you for the good talk. Thank you very much. Absolutely a pleasure, Matthias. Thank you very much. Thank you. Speak soon. Speak soon. Bye-bye.