Everyone's talking about AI agents - vendors are promising revolution, your competitors claim they're already using them, and your team is asking for budget. But what's actually delivering value today versus what's still marketing hype? In this session, I'll share our honest experience deploying AI agents in production environments - including the wins, the failures, and the uncomfortable truths nobody mentions in sales pitches. No vendor slides, no theoretical promises - just real results from real implementations. Drawing from our work at run.events, we'll examine where AI agents are genuinely transforming operations: **Where agents are actually delivering ROI:** - Intelligent matching and recommendations that increase engagement (and revenue) - Workflow automation that handles complexity, not just repetitive tasks - Dynamic personalization that scales without scaling headcount - Resource optimization that adapts faster than any manual process **What we'll cover:** - The real cost of implementation - time, talent, and infrastructure (it's more than the license fee) - How to evaluate agent solutions: questions your vendors hope you won't ask - Change management realities: why the technology is often the easy part - Where agents fail spectacularly - and how to avoid expensive lessons we've already learned **You'll leave knowing:** - Which use cases justify investment today vs. which need more maturity - How to set realistic expectations with stakeholders - Red flags that signal an agent project is heading toward failure - A practical framework for evaluating AI agent opportunities in your organization Whether you're approving budgets, leading digital transformation, or simply trying to separate signal from noise - this session will give you the clarity to make informed decisions about AI agents. No hype, no hand-waving, just what it really takes to deploy AI that your organization will thank you for.

Everyone's talking about AI agents these days - but what are they actually good for? In this architecture and development session, I'll cut through the hype and show you how we're building intelligent agents that solve real business problems (not just generate fancy demos 😉). I'll take you through our journey implementing Microsoft Agent Framework and LLMs to develop agents that actually deliver value. No theoretical hand-waving - we'll dive deep into practical scenarios where agents excel: understanding context, making intelligent decisions, and adapting to user needs in real-time. Through concrete examples from various domains (including our work in event tech with run.events), we'll explore where agents truly shine: * Context-aware recommendation systems that users actually appreciate * Intelligent workflow automation that goes beyond simple if-then rules * Dynamic personalization that adapts to individual patterns * Smart scheduling and resource allocation We'll cover: * The tech stack we chose and why (including spectacular failures along the way) * Architecture patterns for building maintainable agent systems * When to use agents vs. traditional approaches (spoiler: agents aren't always the answer) * Live coding examples - because theory is nice, but seeing is believing You'll learn the difference between "AI agents" as a buzzword and agents as practical problem-solvers. Whether you're writing code, designing systems, or evaluating AI solutions for your organization, you'll walk away knowing exactly when agents make sense - and more importantly, when they don't. Join me for a no-nonsense exploration of what it really takes to build AI agents that users will thank you for, not just tolerate.

Can Europe AI? You have probably heard about "Le Chat", and AI chat similar in its functionality to ChatGPT or Claude.ai. But, Mistral AI, French company behind "Le Chat" has much more to offer, and at the moment, together with German DeepL, it is the best what Europe has to offer in the AI arena. Also, Mistral AI is committed to open-source development and has released several models under open-source licenses, which makes it a competitor not only to Open AI and Anthropic, but also to DeepSeek and Meta's Llama. This will be a deep dive session into Mistral: it's capabilities, it's lovely French quirks, and how can you make the use of it. In this session we will discuss: - Mistral AI basics - Hosting and consuming Mistral AI (Mistral APIs, Azure AI, self-hosting) - Using Microsoft tech stack (Semantic Kernel and Microsoft AI Extensions) to work with Mistral AI - Real life example of using Mistral AI within a product development on the example of run.events À bientôt!

The world of AI is no longer reserved for cloud providers — some AI models can run on your own machines, with surprisingly good results. But how far can you really go? In this session, we'll explore what you can achieve with solutions like Ollama and Stable Diffusion, right on your own hardware — whether that's a high-end laptop, a beefy desktop, or a small local server. We’ll cut through the hype and focus on the real-world capabilities, limitations, and practical considerations. What models can you run? How much performance do you need? What use cases make sense? Where does local AI shine — and where does it fall short? Along the way, we'll dive into concrete examples how we utilized Ollama for run.events, using Semantic Kernel, Microsoft's framework for building AI-driven applications, to show how you can combine local models with powerful orchestration to create real, working solutions. If you're curious about the real possibilities of self-hosted AI — and what’s just wishful thinking — this is the session for you.

We all know the drill: event organizers agree with the venues on the expo plan, then they send it to potential sponsors and exhibitors who decide if they are going to participate and choose their spots based on that plan. They purchase booth furniture and equipment from the booth builder, and then at the event, there is a surprise moment to see if everything looks the way it is supposed to and if the chosen spot was really good. But what if sponsors and exhibitors could walk through and around their booth months before the event happens? What if they could configure their booth based on the impressions from that walk-through and immediately place their order with the booth builder? What if event organizers could then walk through the expo where all booth configurations would be immediately applied, all of this months before the event? This session will talk about how event organizers can effectively use Metaverse to automate and visualize the process of booth selection, planning, and ordering, as a collaborative process between organizers, exhibitors, and booth builders. You will see how the legendary Rheingoldhalle in Mainz, Germany, has been recreated in the Metaverse to help organizers plan the event expo, and sponsors and exhibitors to configure, visualize, and order their booths. This session whil shed light on how Metaverse can *really* be used to the benefits of the event organizers, how this particular use case has been implemented, and what the experience was like for both the event organizers and the sponsors and exhibitors.