Transcript: In Control: AI in the Event Industry Straight Answers to Difficult Questions

yes hello everyone um after quite some time uh it's again me hosting this run events in control webinar today totally proud to have lee stott as a guest hey lee hi how are you hi everyone so um Lee is going to tell you a bit more about himself in the next minutes. But it's suffice to say that we have a very hot topic today that nobody wants to talk about. AI, artificial intelligence in the events industry. I'm coming back to that why is this a hot topic that nobody wants to talk about in a few minutes. But I thought, okay, who would be the best person to talk about this, and who better than Lee Stott from Microsoft, who is, he will tell a bit more about what he is doing, He is in charge of a group of people who are actually doing a lot of AI work on Microsoft side, and he's going to help me and all of us to understand a bit more how this great new technology, which is not going away, spoiler alert, AI is here to stay for a long time, can help all of us in making our event, planning our event, managing, delivering, and analyzing our event, and what are the possibilities there. So I'm totally looking forward to that. But before I pass towards the Lee, I'm just going to say I have been in London the last month together with my colleagues at the International Conference. It was a great show by the day, as always. and I have been participating in few panels regarding ai and what kind of stunned me is that people who were present in those panels either as a panelist or as an audience they're actually there to complain. They saw this as an opportunity to vent out about this evil, evil AI who is going to totally change all our worlds, all our lives, maybe to leave us without a job. Spoiler alert, it will not. I have tried few times to intervene with the constructive suggestions how we can turn this everything to our advantage because advantages are many and this is why I wanted to have Lee here. But the prevailing atmosphere, the prevailing mood in the event professionals community is still pretty negative towards AI, which is actually pretty much surprising me. So this is why we are here. I will be speaking from the position of somebody who is working with AI for many years and still is the person whom I turn to when I have questions and when I need to know something. But also some, but also the CEO of the Run Events, the event management platform, which does embrace fully AI and builds lots of new features and improving the existing features on top of AI. A week after international conflicts, there was a Microsoft AI tour at the very same location in the Excel London, where Lee was one of the speakers. Lee, welcome to this webinar. Thanks, Ades. Hi, everyone. I'm Lee Stott. I'm a Principal Cloud Advocate Manager at Microsoft. I'm based in the UK, but I have a global team. My team ranges from China all the way through to the US, and we're focused on what we call AI calling. So what that actually means is what is the new Microsoft AI Foundry product. So this is a comprehensive suite of tools and services that cover both standard AI services. So anybody who's used cognitive services in the past, that's now part of what we call AI services. And of course, the new modern AI services from large language models, small language models like the five, four series. And of course, agents, you know, it's twenty twenty five. It's a year of agents. And I think. Thank you. You know, I've worked in the industry for over twenty years in IT. And previously, I also had some experience in events businesses. So I did run a team at one of the universities in the UK running all their events. So I know the events business quite well. So this is very close to the interests I have. And I'm here really to help support and dispel some of those technology myths and hopefully address some of the questions around just because we can do it. Should we do it? So the responsible AI aspects, and also where the opportunities are, as Adi said, there's so many opportunities. So great to be here. But Lee, before we actually start, just tell me, are the people who are with us today, who are listening, I said, will they be forced to learn Azure AI Foundry? Will they have to know what's five, four? No, no, there's no need to know anything about that. We're going to go very back to the basics. I'm going to go back to, you know, you're an event organizer, you're an event attendee. What can AI do for me to help my customers or to help my experience at that event? So, yeah, we're not going to go deep in any technology. I've got no slides. I've got no demos. I told Lee just in preparation, just a few minutes ago in preparation for this webinar, it's really my goal for this webinar is actually have a few is to remove fear because I see a lot of fear which is caused by just not knowing the subject matter what is going on to remove to remove the fear and to try to talk about possibilities and opportunities and chances that we all have got because I'm not only running that ceo I'm also as lee knows very much because he's helping me a lot in that regard I'm the main showrunner for the ecs it's a huge technology event in germany focused on the microsoft technologies but we are we have usually like three thousand plus people uh year after year so we are working very closely with lee and his team who are all going to be there So I know also all that from two angles, somebody who is actually in the event business himself for the past twelve years and somebody who is technologist in a way leading the development of one of the leading event management platforms in the world. Lee, are the people who are listening us going to lose their jobs to AI? No, no one is going to lose their jobs to AI. It's about how you become effective with AI, because honestly, it will make you much more productive, a lot more effective, and increase profits and operating. That's the whole goal of why AI is being invested in and being developed. But didn't Bill Gates say, what, three, four days ago, that in ten years, most of the things we as humanity do right now that humans will not be needed for this job. Spoiler alert, I know that this sentence has been misinterpreted for the past seven days and everybody is misinterpreting it really, but let's debunk that as a first myth. What did Bill think about that? Yes, I think it's about how people become more effective in the roles they're undertaking. So if something today can take three hours, which can be reduced to twenty minutes, it's not reducing that task. That task still needs to be completed. Now, at the moment, AI might not be able to do one hundred percent of the task. It might be able to do eight percent of the task. And you'll hear a lot of terminology of human in the loop. And what that really means is that, you know, there should be a human to validate what the AI is undertaking and what the AI is completing. So, you know, people will become much more effective, much more efficient. And I think that's what we've already seen today in, you know, Gartner reports, et cetera, is that organizations become more effective and more efficient by the use of AIs. Now, just because AI is there, it doesn't mean you have to utilize it. It's how you utilize it in an effective way. And I think what we're going to talk about is some of the scenarios where you can effectively utilize AI. And there's some scenarios where we're just not there yet. So it's something that's still very experimental. Right. You basically already spoiled my next question to you. Is AI going to be around the next five months, five years, fifty years? I think it is. I think, you know, the problem we've just got at the moment is the velocity of change. I think, you know, everybody's seen new models where New models were monthly, now they're daily. And now we have things like the model context protocol. So this is the big trend over the last week is the ability to use agents to perform tasks in an autonomous way. Before, we called it robotic process augmentation, so RPA, where you had a task, so let's take this back to the basics you've got a workflow and you want the workflow to be processed by the agent solution and before there was lots of manual steps that needed to be undertaken with the new model context protocol you now can add additional features like being able to call internet browsers or being able to look at specific data sets or do specific tasks. So it's making that AI experience much more simplistic from an end users perspective, but from an infrastructure provider, technology providers experience, it's becoming much more difficult because you've got to consider security. You have to consider responsible AI. You have to consider that human in the loop aspect. So, you know, things are becoming easier for consumers But I'd say from a technology perspective, becoming more complex. But from a user's and an operator's perspective, I think this is where the opportunities lie. And again, you know, when you look at the events industry, there's so many things today which just take so much time. I can definitely see this. We are going up. just for everybody else we actually had a script that uh what we were going to use for this webinar but I think we threw that script only like four minutes ago I'll be looking at it for from time to time but we are going to just freestyle uh to the end and uh use it for us I had conversation yesterday with my team ecs team we are doing this and I told them some of the things that I'm planning that we are planning to change in the next year in matters in regard to operations and sales, especially sales outreach. What just Lee said, try to imagine those things, you still be doing them and overseeing them and looking at going in the right direction, but instead of two hundred emails, which all of you, all of us write, you reduce that to five to six emails and actually have time to do creative stuff that you need to do for your events. Think about reaching way more potential attendees, way more of potential sponsors. Think about all the other stuff where you can in the area of operations and sales, which takes, I can be speaking about us, but I know it's pretty much the same for everybody, which takes up to thirty to forty percent of our manual labor time in organizing the event. If we can trustfully delegate those parts about AI, because nobody could convince me that anybody could like writing emails eight hours a day. I just don't believe that. I think there's lots of things where, you know, so let's break this down into like the simplistics of how people use AI today. So, you know, we've got really two different camps. We have what we call like co-pilots. So, you know, you're in Microsoft Office, you write a document and you can then say like, co-pilot, rewrite this document, summarize this document, rewrite this document more professionally. So again, you know, think about this from an events industry perspective, you know, events are really driven by websites and promotion and marketing tools. So, you know, you write your copy and before you had to think about, you know, is the copy good? Is it bad? Is it spelled? Is it translated? Do I need it localized? that copilot feature is now there to be able to literally do that in a few seconds. So you can give a prompt as translate this from English to Chinese, and it will translate your entire document to Chinese. So, you know, think about the process and time that was involved in doing that previously. It was like, do I know anybody speaks Chinese organization number one? Do I find a localization partner, number two? How long do they need to localize it, send it back to me to validate it, to proof it? Now, machine-based translation isn't one hundred percent effective, but it's like ninety-nine, ninety-eight percent effective. So for most use cases, it's a great starting point. So Copilot has been your starting point to get started. Another great example is like, you know, you want to create a brief or an abstract or a marketing quote that you want on a banner. You can give it a simple prompt of this is my event. So you can give it the agenda of the event. Create me a marketing summary. SEO optimize that marketing event. And again, it will do that. So that's that's what we call the copilot experience. And that, again, can be from create me a PowerPoint deck or a template and it will do it. summarize a document, or even look at my data that I've captured around this event in an Excel spreadsheet or in a database and provide me some insights. And it will provide you insights, draw your dashboards, draw your charts. So I become much more effective. And this is where we're starting to see this leveling of the playing field. You know, an organization before may have had people. Now a very small startup with three people can be as competitive because they're using AI to support that business decision making. So that's where I see like the co-pilot part. Then we look at, you know, what we call the development aspect. So People like you, Addis, you're running an events organization. You're probably going to use the co-pilot from your organizational perspective, so your teams and your marketing teams. They might even be doing really cool things like Adobe Creative and creating videos with AI or banners with AI or artwork with AI. Again, we just launched GPT-for-O image where now you can say write this text on an image and it's pretty good. It's very impressive. That's not getting rid of the graphic designers. It's just empowering the graphic designers to be much more capable and deliver content on time. Now, people will say, yeah, but you don't need those skills. You do need those skills to make sure it's fundamentally correct. It's going back to the developer story. Developers aren't going to go away because somebody has to be the expert. But from that event side, it's how do you build those solutions which then really add value to the partners? So people you want for sponsorship, how does it add value to the attendees of that event? I want to meet people unlike me at that event, for example. I want to meet people interested in this specific topic or this specific product. AI can be a great enhancer to that experience as an event attendee. So if I make... step back but build on what you uh said it's like imagine hundred twenty years ago and the car industry people are afraid people like what are those things going to the streets driving to the streets will I need to learn how to drive a car yes you will if you want to drive if you want to move faster you will need to learn how to drive a car but will you need to learn how to build a car no That is not your job. You will never need to know unless you really want to know how to build a car. You will never need to learn how to build a car, but you will learn how to drive a car the same way you learned how to use a computer, the same ways how you learned. So it is just a new tool in a tool set, but tool that can shift the productivity, in my opinion, much faster and a much higher degree than we could imagine only a few years ago. so uh right we have one of the questions already here if ai is handling the translation that's great but it's important that the person who knows chinese still reviews that's the truth while the ai solution is in the testing phase the human role is crucial to supervise the ai's output I'll be interested to hear your uh opinion it was from tom tom I couldn't agree more I will tell you just an anecdote from yesterday We have released yesterday the translation of RunEvents, of the event management platform that we are building, to Dutch. RunEvents was effectively translated into Dutch by Mistral, and the translation was corrected by FiveFour and by DeepSeek. We actually used three different AIs to work with each other or against each other, however we want, and right now it's being... corrected or approved by my good friend and colleague thomas fortin who is a native dutch speaker so uh as the tom gavazzi just said yes you need the human element it's just the question did we need two months to space it or two days this is actually an awesome example well or twenty minutes so I've just put a link which anto share our producer will share in the chat we we've actually open sourced a solution to do exactly that ai translations And what we're trying to really do here is do exactly what we've said, is reduce that time to market. We want to produce GitHub reviews. which are localized as many languages. Now, having a repo which is ninety-eight percent and then we can take an issue back from users where they find typos or, you know, issues in the formatting of that repo is much more effective to us because then we can, you know, it's reinforcement learning. We're actually using you as users to reinforce that learning, which we then can train the model, build a better model. And in this case, we're taking like, you know, GitHub translations to twenty minutes and this will translate all your images on your website all the text in the website to whatever language you want and we split twenty languages today and that is just making it making it more effective so that's a great example of using ai to enhance your business opportunity you know you want to go to more markets You know, you're putting a disclaimer in that to say this has been AI translated. You're adding that human in the loop aspect. You know, we could have human in the loop from our perspective, but because we want to do this as a community project, we want the community involved. So great example. So, but we now are deep diving in what I call... chat use case which we definitely have in the which is which are usually focused on organizer value which you as event organizers uh have definitely benefit from ai in order to make your lives easier and uh no you are not going to be replaced by ai but you are going to be really faster if you learn if you open up to learn uh to drive that car but this is I definitely wouldn't stop there And we will go into each of those aspects. There are huge areas where AI could be used for event personalization. Lee just mentioned, I want to meet people with similar interests. I want to meet people with opposite interests. I want to meet people who have nothing to do with what I do, but I'm interested maybe in that focus. So tell me whom can I meet at this event that somehow fits and suits my expectations from this event. This is definitely an AI-powered matchmaking, which is a big deal and which is possible, but there are preconditions for that, which I'm going to talk about in a second. There's also then what we call internally product spotting. Now, if you come to our ECS event, it's a small expo with seventy-five booths. We have been at international conflicts which had over three or four hundred booths. International conflicts is for two days. There is no way you can have three or four hundred conversations in two days. If you do, you are a legend. But I know that I wouldn't be able to do that. So how do you spot who is interesting for you? How do you spot what products are interesting for you? So there is a whole personalization features in AI used for product spotting. Also vice versa for the exhibitors. We speak, you know that exhibitors will always go to do the outreach to the audience to talk to them. But who should they outreach to? Do they outreach to all six thousand people who are international contacts? not viable it's it's difficult it's difficult to do something like that so also the other uh the on the opposite side we do have a buyer spotting uh of course the topics of gdpr and involvement and agreements and everything needs to be done and we have this topic inside our scripts for today but there are definitely things which can be which can be uh which can be do there also there many open possibilities for AI augmented marketing and AI augmented sales. We have scratched a surface with that and it's already giving huge results. I would also like today to take a few ideas, to talk a few minutes about that. This is where the age of agents is coming. sales agent zero zero seven for this actually trying to find your next exhibit or trying to find your next attendee and to persuade them to actually come come at your event And on the end, the big part is analysis and compressive data analysis. What did work with your event? What didn't work with your event? What do we change for the next one? Also, if you go from year to year, the analysis, who did we lose? Why did we lose them? Who did we win? Why did we win them? And this kind of stories which were possible without ai yes but again would take much more time and much more manual work than we can we can able uh to uh to uh do the to uh do that now we see the question so let's let's go to tom's question because I think it's really interesting about this you know the whole psychology of individuals at an event and I think you know again this is where ai really gives you the opportunity to personalize I call this personalization so as an event attendee I can build my profile and if I'm extroverted you know I could even have a tag that says you know I'm an extrovert I like meeting and networking you know so number one you tag in your own data You know, data, we all know data is gold. So the more opportunities that you have of personalization to your attendees, the more the attendee feels part of that event, the more data insights you have of that person attending that event. So again, think about the different ways that people can now contact. So I might be completely introverted and I don't want to speak to anyone during the event. I just want to go to my sessions and leave. But I might have a preference of saying, contact me on LinkedIn. You know, I'm quite happy to take an email or a LinkedIn message, but I don't want to speak to you and I don't want to meet you. So until I know who you are. So, you know, think about that. That's a really simple technology solution to fix is when you build your personal profile, your preference contact method is LinkedIn. So when you're doing your networking, it's like, you know, Lee, I'm really interested in speaking to you and you would get a LinkedIn message. So again, I think there's lots of ways we can easily deal with different types of personas and different types of individuals. And that personalization is key. And having the control to the individual is really important to that. Totally. All our questions or our next questions are basically starting to revolve around this networking is somehow crystallizing from the question around networking and B to B. and but also privacy for any kind of matchmaking buyer spotting product spotting in ai this process needs to be voluntarily for all sides everybody needs to be really easy to opt in and again opt out again and as lee said to choose the prior to preferred contact methods this is what we are doing in running bands. This is how we are building it. But I think it's not only GDPR that we are GDPR compliant. It's also common sense. As Lee said, there are different people with different preferences and they all need to be respected. They all need to be accounted for. If I don't want to participate in any kind of matchmaking, AI support or not AI support it, I need to be able to participate. It's as simple as that. That doesn't change with AI. That maybe just gets amplified with AI a bit. But the whole idea of event personalization can be taken many steps ahead doing this because what would be a good example of using AI in high value fundraising events. Thank you, Anto, you were just amplifying the message. Not only in fundraising events, but especially this, you need to match people who are raising the funds with the potential donors, with the potential funders. For AI, if you have those profiles complete, again, for the people voluntarily completing their profiles, the it is relatively easy from a technology perspective not to go down deep down into technology it's relatively easy to actually figure out the perfect pairs for them yeah let me explain this a little bit yeah so so there's a a process called uh so it's called rag so r-a-g and basically what you're doing there is you're augmenting the actual data so you have your events database management system which says you're a fundraising company you're looking at people's fundraise for charities etc you will have in your system a profile so you're like looking for this type of organization this type of donor so they're like your base level data what you can then do is take your sponsors data so again everybody creates bios or abstracts you take that data and then you say to the the llm system match me the donors from their profiles against the items that we have for auction, for example. And it will just literally go through, read all their profiles, understand what they're interested in, and then match them to some of your items that you're fundraising against. So immediately you've created that network of opportunity. You've seen these are the potentials, these are the potential matches. So then you can think about, OK, how do I then do demand gen to those people and say, oh, we see that you're really interested in X. Did you know auction three thirty four is Y? You know, would you like to bid on it? So that's a great example of how you've used it. to go to market as you go to market with your ai solution again think about this from a user's perspective so when you you know want to introduce that user to those potential items as well as you can do the other way around so you can take the individual's profiles and then say oh we've already got these items available are you interested so it's exactly like what amazon does today you know when you go to the amazon store and it says oh we've seen you purchase this would you like this it's just an ai algorithm that is happening in the background looking at your previous purchases your spending patterns your profile and then saying we would like you to to have these items so uh pretty much the same question just your face from daniel uh who is asking uh what are some practical ways to leverage ai uh for networking and sponsorship in the b to b events so Let's take even the ECS with only seventy five sponsors. If you are a technologist and you are interested in particular governance products, those who are in technology will know what they are. And there are seventy-five booths. Are you going to go from booth to booth trying to spot on the back walls and the branding, would they be into that industry? Would they have something like that? Or are you going to ask them one by one, hey, do you have solutions for data governance or information governance with you? Or would it be cool if you would get a push notification to say, hey, Lee, I see you are into governance. And we speak about push notification to your event mobile app. I think if you are into searching for a product which will help you with that, you should look at the companies A, B, C, and D, which are on the boots, X, Y, Z, and W. So this is what personalization can do to help. And this is really, this is, I did not just invent the case. This is the case at our own event, and this is what we are doing. to help attendees to do exactly this. This is a real case example of what's going to happen in two months. And this is where we as run events put a lot of effort into enabling exactly those capacities into our platform to do with that. yes so thank you brie for just confirming rag rag is retrieval augmented generation so it's basically adding additional data to your your models it's a very simplistic way of doing it gerrit's question is brilliant and this is something me and addis we're actually speaking about so he's saying you know the events uh the coffee line we've all been to events we've all seen that coffee line You know, all the catering queue and how do you bring gamification to this? And, you know, one of the key areas now is augmented reality. And, you know, you can actually use, not necessarily augmented, you know, we're not going to be walking around with devices on, but we all carry mobile phones. Mobile phones have location-based data. So you can start to think about how do you add that gamification into your event system. If you have an event app which has location data enabled and the user has subscribed to enable their location, You can build hotspots in your event space. So when people in the expo space, when they're in the coffee line, do you give them broadcast information? Do you create that gamified aspect to where the user is in the experience? shameless plug what we are doing we do have gamification features in the run events as a platform which actually uh bring you to go to the booths and to have coffees and a lot of our a lot of sponsors can have their own coffee so to get people together but there's one cool thing we do with ai when you network with our mobile app from run events and I scan lee's badge get a short info if lee opted in then again really important always to say I get information about who is lee that he's working for microsoft he's into ai he's leading the team of the people doing that and I get the perfect opener for lee uh perfect opener for lee might be hey lee can you talk to me about uh rag and he then you will need to listen to early uh for about two days talking about that personal things it might say you know I'm really interested in camping or whatever yeah so yeah all good so exactly what Garrett says conversational and I wouldn't really put only introvert extroverts it's a bit too binary for me I like much more what you said Lee I might be introvert in a way that I don't want to talk to two hundred people today But I'm okay with five or ten people. The rest, let's do the emails, let's do the LinkedIn. So there's a lot of nuances between I'm an introvert. I don't want to talk to anybody. And I'm an extrovert. I want to talk to everybody. I mean, there's a whole palette of gray in between that black and white or however we want to put it. So I think there's a lot of possibility that we actually improve that kind of matchmaking that kind that kind of uh that kind of things which are which are going so I think you know one of the key things we need to cover is data quality I just I know it's like don't even start me with that now you'll hear me talking about two days about data quality yeah But I think that's something as an event organizer, you really need to think about your data estate. Is it multiple spreadsheets? Is it a database? Because honestly, it's an age-old story of what we call garbage in, garbage out. If your data isn't right, your outcomes are not going to be right so I think that's the number one thing to consider yeah again you know just because we can build it should we build it so that's the human consideration aspect you know does it add any value because there's lots of things that you could do like I said before you know we all do we all want to be walking around with apple provisions on not sure it's we're really there yet and that's a huge investment to the event space But even without AI, let's just leave the AI for the moment, as I have been talking to hundreds, even thousands of event organizers for the past years about topic of data silos and data connectivity, because it's a big part. The data... It's for event organizers, for event professionals. Data is the highest value, highest good they have. With data from this year, they can sell out events the next year. Or with the exhibitor data, sponsor data from this year, they can find more sponsors the next year. And then I so often hear the stories, oh, we use ticketing app X to sell the tickets because they really have this one great feature for ticket selling. Or then for sponsors, we usually mainly Excel, maybe some of more advanced ones. I will tell you some CRM in the back end that we have. Or for speakers, maybe we just use email. We ask them for the topic abstracts and we put the topic abstracts from the email to the webpage and there's nothing about that. And for the mobile app, we use a fifth mobile app where we import the data from A and B, and sometimes it works, sometimes it doesn't. For example, international conflicts, which you would expect them to do, my batch was wrong all the time because they didn't have a real-time sync. They had five days of import. Doesn't matter. And I always ask, okay, how do you do analysis after the event? And they're so proud that they managed to use, I don't know, some workflow tool to interconnect everything to whatever. It's like, how do you do analysis? And then you see here, blank space, analysis what? Well, analysis, how do you know what worked, what didn't work? Also for marketing or whatever. Well, we didn't think about that. Now, you could get away with that in a way, but if you want to use AI, AI needs to have access to that data in order to do anything. Let's just think about product spotting. You need to have info about user, attendee profiles. Who are the people? Do they want to be contacted by LinkedIn? Are they into governance or are they into cats or are they into camping or what are they into? You need to have your ticket information. What are they visiting? What part are they not visiting? You need to have information about your sponsors and exhibitors, about their booths, about their products, and about their own contact methods. And, of course, you need to have a mobile app which connects to both A and B in order to connect everything. Now, if you have this in three different systems, forget it. It is just not going to work. It will not. In RunEvents, we have something which we call event intelligence graph. It's basically a data network which connects dots between speakers, sponsors, attendees, products, sessions, visits. For example, also the connections. If I am connected with Lee, probably Lee and me or three or four people from Lee's team, probably people of the similar profile would also be my potential next connection. So you also need to have information about networking information, who is connecting with whom. I told you I could talk about this for days. The thing is, in order to make anything of this work, you need to have normalized data that AI will have access to. If you proudly put your ticketing on A, your marketing on B, your sponsors on C, and your speakers on D system, nice. You have done a lot of work on integration of that with very limited return value. And that's something that you need. And let's not even start about analytics later to see which sessions work. How can we help that for the next year to have more of those sessions? Which attendee profiles have been visiting which sessions? And to try to find correlations, who is listening to what, who is attending which sessions? Which attendee profiles have been visiting which sponsors, et cetera, et cetera. All this analysis and intelligence work, and we still speak about artificial intelligence here, need to have access to that data in order that we do anything. So we will be happy if it's run events. It doesn't have to be run events. But please, if you want to make that leap step forward, get away from the isolated silo systems. They are not going to bring you anywhere. Not in this regard. Sorry. Tom's asked a great question here about the whole issue of hybrid events. So an online event, an in-person event, and how do you keep the event running throughout the year, I think is a really interesting question. I think for lots of people, I think the networking is a key opportunity to that. And again, you've got to think about Where do you want people to go? And again, going back to, you know, will this reduce jobs? No, we won't because you're going to need more staff depending on how you build that network. You know, do you build a network in Discord? Do you build a network in LinkedIn? You know, how do you take that conversation going? And then it also feeds into that conversation. You know, is that conversation going to be human to human? Yeah, you want humans there. but are you going to build an agent or a triage bot to be that first point of contact with that user in that discussion forum so again you know this is where again we're into the world of agencies you can build some really smart intelligent agents today which handle ninety percent of those questions you know so it understands the frequently asked questions it understands know your event registration prices it understands things that were discussed it understands where the content is available to point people at youtube videos or on-demand videos so again this is how you start to build that that further discussion and engagement around those events because most people who attend event want to either network with people after the event or want to re-consume content that they've missed and again going back to that personalization It's about you then providing back to them. Oh, you know, we know you attended the event X on this day. You attended these sessions, but you missed these other sessions which were related. Would you like to watch them? Would you like us to build your personalized playlist? Would you like to have them as a Spotify playlist? Would you like it as a podcast? Would you like it as a YouTube video series? So, again, you're building this new audience, which hopefully, Tom, you know, will sustain people for that year. And it also provides you a way of doing marketing demand gen for next year's event because you can say to people, oh, you attended this session last year. We saw that you attended twenty sessions. You missed these sessions. This year we've got the follow up to these sessions and we've got the new features for these things. And then we are coming again back to the data quality and data normalization for that. If you don't know who has been at that session and how did they rate that session, even if you have a session feedback system in place, we have two questions which are based on what we have been just talking about from Grail and from Bri. Grail is asking, what practical AI tools can we use to assist with registration via CRMs and in-person ticketing batch collection? I can go fully blunt marketing and say, run events. Anto can now type in the website. So we'll be happy to show you, but basically take some, any, I mean, any good event, event platform, which can actually help you to do this, to do, to, uh, where the data will be normalized, will be on the same platform, and which will allow you to connect to the CRMs, if you have got them, because I realize that there is a need always for that. And for everything, registration, badging, sponsors, boots, expo data, attendee data, speaker data, session data. take a look at run events or but we are not the only ones obviously on the market similar from brie yes but how do we get all this data into one place how can you get the teams to accept yeah the same way no governance same way we got the people to accept the cars yes you we could have continued with carriages and we would get from a to b A bit slower, but we would get from A to B with cars. But without moving to cars, we would never be able to open capabilities which we have by reaching larger distances which are more far away. This is... And there is technologies which allow you to bring disparate data into single locations now. So, you know, from a Microsoft perspective, we have things like Microsoft Fabric One Data, One Lake, which allows you to have your disparate data sources, but then bring them into a single database. So everybody's still running like flat file databases pdf files excel files you might have sql database you might have an access database you know yeah you know you've got a disparate estate and you can bring it together all these other technologies like azure search which is used more from that llm perspective you know from the rag perspective because you want to reinforce data into your models and again you can search from across different data services so you know, really it's about the data qualities. You know, if you've got garbage in one point, you're going to get garbage in all points when, yeah, it's like, cause it's, you're mixing it all together. Lee, I have, I have tends to send people who are starting with a separate system, silo system and Excel files, number one Excel files. Somebody tries to start talking about Excel files. I'm paying the stake, the best stake in the town. If you explain me, and you convince me that you cannot miss your Excel files. This is normal human need of trying to feel cozy and I'm in control. Like with my Excel file, I'm in control. You are in control, but you are in control of this one sheet. You are not in control of basically anything else going outside of that sheet. And this is a shift. Bri is right in this one. It is a shift how to convince people But basically, the only way how we convince teams, how we convince people is to tell them, we are missing on opportunities, we are missing on money, we are definitely probably missing on money to be made in the next events, based on the data, which we cannot be using to actually make, make fun, make conclusions and make decisions, data driven decisions for the for the for the next for the next year's for the next edition of our events. Tom's asked a really interesting question. So he says, one of the key events is communication. So communication channels is key. Naturally, most people spend their time in WhatsApp. Yeah, that's fine. Today, people increasingly look to download yet another app to engage. And this is, again, where agents hike so good. That's so good. So there is now a new MCP. So MCP is Model Context Protocol. So MCP TCP IP, the internet is a protocol. It's a standard. That's what it basically is. It's a standard by multiple organizations to get together and agree on a protocol standard for communications. Now, MCP allows you to take agents. So again, you will need your data. You will need your event system. And there is an MCP server now for WhatsApp. So you can say, I want to communicate via WhatsApp to all my attendees. And I'm going to use the MCP server. protocol to do this. And I'm going to engage with my users who are interested in these following areas. So again, going back to that preferences thing about how you want to engage, it's just a tick box. Contact me by WhatsApp. Contact me by Discord. Contact me by LinkedIn. Contact me by X. All you're doing is setting those preferences, but then your system is engaging with those users where they feel most natural, most confident, most secure, most friendly, et cetera. So Anastasia is writing that Germany has the greatest number of chat GPT users. First open AI office in Germany is now open in Munich. We are not that back-forced as some people sometimes claim. RunEvents is the first event platform which fully embraced AI and builds on AI. We are a German company, however, not from Bavaria. We prefer it a bit more in central Germany, but still Germany. So, yeah, Germany is not a bad place to be in this regard right now. Lee, we've got ten more minutes and we've got two important topics which I want to tackle before we leave, which are more on the social segment in part, less on technology, but it needs to be tackled. This one is, yeah, let's speak about that cat in a bag, it's costs. How much does it cost? Tokens. Yeah. dollars euros whatever we want to say yeah so so you know ai isn't a cheap solution you know I think there's a you know if you wanted to go back to that co-pilot solution there's a cost per user per month um again you know that varies depending on the size of your organization but if we just say it's around thirty dollars per user per month for microsoft and three sixty five co-pilot As we said before, though, does that thirty dollars save two hours of work? If it does, I think you've made your money back immediately, you know, to build and implement your own solution. Again, there's a cost. You know, you are paying for that token and a token is basically a set number of characters. So if you break a sentence down into tokens, a token is really that prediction of the next word that is going to be formed, which is then broken down and you pay per token usage. So, you know, the average, there's no real average. That's the problem. It depends on how effective you are with building your AI solutions. You know, if you start at the very, entry level, which we call prompt engineering. So you write a prompt and you get a prompt back, you're paying for the tokens for each sets of prompts. You then move into that rack, so the retrieval augmented generation, where you're adding more data, potentially adds more tokens because you're adding more data, or you move to the level, which is really like expensive, which is fine tuning. But the quality of the model based on a fine-tuned model is excellent because you have fine-tuned that model to that specific event or activity and the data is amazing. So it really depends on how you want to approach this in terms of a technology solution or you work with a partner like RenovEvents who will, you know, you're building on top of their platform, which is having all that customization done. Right, as what you just said, I can tell now from the perspective of RunEvents, because we are your users, we've actually paid you for the services that we offer to our organizers. There are two things. If you use our infrastructure, as Lisa is difficult to say, how much are you going to use our infrastructure? So we are trying to then come up with the price per person per attendee, which is going on there, which is basically estimation. and that we know how much one person will be using AI features, those tokens, at least talking about throughout the event. We are trying to reduce those costs by pre-processing the user data before the event, when it's cheaper, where you don't have real time transactions, but where a lot of data can be pre-processed and then stored and then consumed during the event. That does reduce costs to a large degree another option we can always talk about is basically that you bring your own microsoft azure subscription or open ai subscription to us and we say okay we don't care about your ai costs you use the ai augmented running event features But actually I cost, they are then again paid by you, but to a different payment channel, not to us, but to Microsoft, to AWS or to whoever Microsoft for those things. That's also a possibility which we can talk about. But I'm not going to downplay that there will be costs per user. We basically still talk to each event organizer that comes to us and to try to understand the needs and what they want to use, what part of that can be pre-processed before the event. What part of that needs to be real-time during the event and how are we covering the real-time story? And post-event, it's always post-process. So that part is also cheaper because it can be processed in different times or a different phase. So that part is also coming to the cheaper variant. But basically, we need to talk to you to understand you uh uh what uh what your needs are and then basically to take the units and to estimate the price which is uh which would be we had a question from anna who asked like how was it non-profit run events has strong discounts and we are even for certain type of non-profits we are even for free however we cannot be uh we cannot give ai for free because we are paying it to leave So the company it works for. So those are the, we need to at least cover this part of costs. But then again, Anna talked to us, we will try to find a way to see what works, what doesn't work. The price of AI event assistant depends on the amount of content. Okay, I need to make event assistants. I'm going to tell you something that a lot of people are not going to agree with me, but I don't care right now, right now in this moment. If you put a chat GPT on your website and augment it with some data, you are not doing AI. I mean, you are kind of doing AI, but you are tapping into zero point five percent of possibilities that you have got. So, yes, Tom, I hear you, what you are saying, but I actively decline. Of putting putting putting a tool for calling just splashing an ai chat on the website and calling we've got ai I just actively refuse to to accept that this is our limit with ai which we which we have got uh what here I'm going to do it and that two biggest questions we will be adding to us will there be a guitar performance um sorry I'm sorry, but I might do that later. It depends how much. And what's the name of the cat, Lee? What's the name of the cat? Yeah, this one's Shadow. So this is Shadow. She's black. So she's called Shadow. And this one's called Shade. I don't know where Shade is. Shade's around somewhere. Okay. The last, we've got a few more minutes. The last one is really what I called here security, compliance, responsibility, and trust. And the most important thing, start with this. Don't think about this at the end, number one. In Europe, as you know very well, Lee, because we have been discussing in the past few weeks a lot, we have got something which is called AUAI Act, which is the first legal framework done by any country or association of countries in order to try to not only try, but to kind of guide rail the use of AI that it complies to the ethics standards, to the legal standards and to everything which we know. Basically, you need to be transparent with the people that you are using AI. What kind of AI are you using? What are you using? What are you using it for? and to ask them if they agree their data will be used for that in order if they want to get suggested recommended products or recommended sessions they will need to enter their data as a basis on which we can recommend sessions or products but this all needs double underlined must be transparent clearly communicated and guided by some kind of guidance. If you are in Europe, in European Union, you have this guidance as a law, which is officially working. But even if you are not in EU, you can put and you should put those standards yourself. I mean, it's just basically a matter of nice behavior to tell people what are you doing, especially what are you doing with their data, right? Yeah, it's honestly, you know, responsible AI is so key to everything we do. You know, I started off this by saying, just because we can do it, should we do it? And I think that's a key decision that you need to make with your organizations. You know, as I said, you know, this is about personalization and it's about people opting in and opting out. So you've got things like GDPR, right to forget. You've got the European EU, European AI Act. You've also got other countries and other societies creating AI acts. There's a specific AI act for Californian residents. So, again, you know, you need to think about your audiences and where they are in the world. when they are registering for your event just because your event is in europe what is their local custodianships So again, you know, responsible AI, and again, Microsoft provides you lots of tools around responsible AI. And we really call it observability and evaluations. So that's really like, is the response grounded? Is it coherent? Is it relevant? Does it meet those needs that you're asking it to do? And you give it dummy data to say, this is the types of responses I want. This is the types of responses I get back. And also you need to think about the security aspects. Can someone jailbreak your AI? You know, could they get your event for free? could they ask for your contact database and get all the data for free? So you need to think about how you're going to protect that data that your LLM is based off. So again, if you are fine-tuning the data, if you are doing RAG, make sure there is no PII data in that data you're surfacing. Because once you train your model, that data is embedded in that model. So you will be able to get it back. Right. Good. We are running out of time. We've got one more question, which I'm going to read at the very end. Lee, can event professionals afford to ignore AI? No, I don't think anyone can afford it. No one, whether you're a school child today, whether you're a professional developer or an attendee, no one can afford to avoid it. I think the issue is we all have these devices, don't we? We do. There's more AI in these devices we're using every day. We've all got Siri. We've all got Gemini. We've all got all these co-pilots. A lot of us are now just using it without even thinking about it. It's becoming the norm. And again, for the younger generation, it is the norm. You know, they've grown up with these since they were two months old. the thing is it's like can could be afford to uh could we afford to not to learn drive cars yes we could but we would be limiting ourselves in our daily life I think about others professionals understanding the fundamentals we understand the fundamentals of how these things are built and why they're being built in this way we don't need to understand the complexity of how to do it But if we understand the fundamentals, then it gives us the ideas of it's much more than just a chatbot sat in my app. You know, it can help me do other things. And right. So the last question was from Daniel. If you had to choose one application of AI for B to B events, what would it be? Daniel, really, I'm run events CEO. If I would tell anything else but run events, I would just be plain weird. So I'm just going to say run events, obviously, for the obvious reasons. But because we are also the first one in the event platform that mostly invests in this area, but also in all other aspects of life, which is not only event and event intelligence and everything. There are so many different tools and platforms and technologies which can help you basically do your work. I was just going to say, you know, for personalization, I think you've got event personalization, massive exhibitor value, massive organizer value. And you've also got like empowered sales. You can go to market with specific audiences and also share success stories. That's another key one. Think about how do you build those summaries of the activity? What was the overall event outcome? Use AI to build that for you. Okie dokie. people thank you very much uh I have people telling us that we are that they need to leave so uh we are all going to leave lee thanks a lot I had fun I hope you had at least a bit more of a bit fun as I uh had this was awesome use ai people really it's the closest equivalent is like under twenty years ago people are resisting to try to learn to how to drive cars because uh they didn't think that it would stick and that they will be around. You don't need to learn how to build cars. That's why we have Lee here or myself or other people, but learn how to drive cars and try to imagine how far can those cars can take you, what new things you can do and see when having the car. versus not having the car. So the same is with AI, but amplified and amplified by lot. I think, Lee, that you might agree with that statement. Yep. Thank you, everyone. It was fun. It was cool. And see you in the next InControl webinar, which we are going to announce soon enough with another great topic like this one. Lee, thanks. Thank you, everyone. Bye.