/Events /AgentCon 2025 - Milan
About
Join us for the AI Agents World Tour, a global series of one-day conferences designed exclusively for developers building the future with AI agents.
From San Francisco to Singapore, we're bringing together leading engineers, researchers, and creators to explore the cutting edge of AI agent design, deployment, and integration. Whether you're building intelligent assistants, autonomous systems, or next-gen developer tools, this event is your fast track to practical knowledge, hands-on demos, and real-world insights.
What to Expect:
- Deep-Dive Talks from AI pioneers and industry leaders
- Technical Workshops on building, deploying, and scaling agents
- Live Demos of powerful open-source frameworks and tools
- Networking with a global community of builders and innovators
This isn't just another AI event — it’s where developers meet to talk about real code.
Ready to build the future? Join us in Milan.
📅 AgentCon 2025 Milan – Agenda
Mattina
09:00 – 09:30
Registrazione e welcome coffee09:30 – 09:45
Welcome in the world of AI Agents
Alessandro Geraldi (CEO, Impresoft Group)09:50 – 10:35
Keynote – Beyond models: the era of AI Agents in Azure (in inglese)
Dona Sarkar (Chief Troublemaker, Microsoft Corp)
Carlotta Castelluccio (Senior AI Advocate, Microsoft Corp)10:40 – 11:20
Costruire sistemi multi-agente efficaci con Semantic Kernel: tecnologie e sfide
Anna Fabris (Senior Data Scientist, Impresoft 4ward)
Matteo Pagani (Senior CSA, Microsoft Corp)11:25 – 11:55
Tips and Tricks of Azure AI Foundry
Guenda Sciancalepore (CSA, Microsoft Italy)
Roberta Bruno (CSA, Microsoft Italy)12:00 – 12:30
Model Context Protocol: Principles and Practice
Emanuele Fabbiani (Professor Università Cattolica)12:30 – 13:30
Networking Lunch
Pomeriggio
13:30 – 14:00
PolyGloT: Personalized and Gamified Learning Paths with Generative AI
Antonio Bucchiarone (Associate Professor, Università degli Studi dell'Aquila)14:05 – 14:35
Automated Design of Agentic Systems: a new frontier in AI architecture
Matteo Conti (Lead AI Architect, Impresoft 4ward)14:40 – 15:10
AI Agents for Quality Management – Lessons from the Field
Matteo Giunchedi (CEO, Liv4)15:15 – 15:45
Espandere Microsoft Security Copilot con agenti AI
Mario Serra (MW e Security Specialist, Overnet)
Michele Sensalari (CTO, Overnet)15:45 – 16:15
Coffee Break16:15 – 16:45
Agentic AI with Local LLMs/SLMs: Why Not?
Stefano Demiliani (Microsoft MVP - Microsoft MCT Regional Lead for Italy)16:50– 17:20
Agenti Intelligenti e Dati Aziendali: Innovazione con Teams, Fabric e Copilot Studio
Marco Caruso (CTO, Lobra Futura)
Sessions & tracks
Beyond models: the era of AI Agents in Azure
You've heard about AI. You're heard about Gen AI. And NOW you've likely heard about AI Agents! In this keynote, we'll explore the world of AI Agents, analyze how they work and showcase the Azure services that enable the creation of a custom agentic solution. We’ll also walk through a practical use case demonstrating this technology in action.
Carlotta Castelluccio is a Cloud Advocate at Microsoft, focused on Machine Learning and AI and based in Italy. Her mission as a Cloud Advocate is to help every student, developer or startup founder succeed with AI, by building innovative solutions responsibly. To achieve this goal, she develops technical content, and she hosts skilling sessions, enabling her audience to take the most out of AI technologies. She also engages with Microsoft communities and ensures they have an impact on Microsoft products’ evolutions and refinements.
Costruire sistemi multi-agente efficaci con LLM: tecnologie e sfide
Questa sessione esplora le tecnologie fondamentali per sviluppare sistemi multi-agente basati su Large Language Model. Verranno approfonditi aspetti cruciali come Architettura, Memoria, Pianificazione e Framework, analizzando i progressi recenti, le sfide attuali e proponendo soluzioni pratiche per migliorare collaborazione, adattabilità e resilienza di questi sistemi complessi.
Ho conseguito una laurea triennale in Informatica e successivamente una laurea magistrale in Intelligenza Artificiale. Attualmente lavoro come Machine Learning Engineer a Impresoft 4ward, applicando le mie competenze per sviluppare soluzioni innovative nel campo dell’AI.
Matteo is a Cloud Solution Architect in the Modern Work - AI, Apps & Ecosystem team in Microsoft. In his role, he works with global companies to build innovative projects for the Microsoft 365 ecosystem, based on the latest innovations in the AI & applications development space. He has a strong passion for knowledge sharing and he supports the developer community by writing articles, blog posts and books and by speaking at conferences all around the world. Before joining Microsoft, he was a Microsoft MVP in the Windows Development category and a Nokia Developer Champion for almost 5 years.
Tip and tricks of Azure AI Foundry
How can you get started and use Azure AI Foundry to develop, build and manage your agentic system. During this session we will show how Foundry works and all the feature that make it quick and easy to build your first agent.
I work as a Cloud Solution Architect specialized on AI & Machine Learning at Microsoft, where I help partners developing innovative and AI infused software and solution on Azure. Previously I spent 3 years at Snowflake, working as a Field CTO for AI and Machine learning. I graduated in computer science and specialized in applied cognitive neuroscience. I love to talk about tech, AI, gaming, brain and how we can make the tech world a much more inclusive and diverse place.
I'm a Cloud Solution Architect at Microsoft Italy with a focus on Azure AI and Machine Learning services. I have a strong foundation in statistics and computer science, and I am very enthusiastic about how technology can create positive change.
Model Context Protocol: Principles and Practice
Large‑language‑model agents are only as useful as the context and tools they can reach.
Anthropic’s Model Context Protocol (MCP) proposes a universal, bidirectional interface that turns every external system—SQL databases, Slack, Git, web browsers, even your local file‑system—into first‑class “context providers.”
In just 30 minutes we’ll step from high‑level buzzwords to hands‑on engineering details:
- How MCP’s JSON‑RPC message format, streaming channels, and version‑negotiation work under the hood.
- Why per‑tool sandboxing via isolated client processes hardens security (and what happens when an LLM tries
rm ‑rf /). - Techniques for hierarchical context retrieval that stretch a model’s effective window beyond token limits.
- Real‑world patterns for accessing multiple tools—Postgres, Slack, GitHub—and plugging MCP into GenAI applications.
Expect code snippets and lessons from early adoption.
You’ll leave ready to wire your own services into any MCP‑aware model and level‑up your GenAI applications—without the N×M integration nightmare.
Schedule
0 – 5 mins: The integration headache — why function‑calling alone can’t scale.
5 – 15 mins: Under the Hood — MCP architecture, JSON‑RPC anatomy, bidirectional streaming & security sandboxing.
15 – 25 mins: Orchestrating Tools — patterns and multi‑tool workflows, with code examples.
25 – 30 mins: Ecosystem & Q/A — pre‑built servers, SDKs, when MCP is overkill.
Emanuele is an engineer, researcher, and entrepreneur with a passion for artificial intelligence. He earned his PhD by exploring time series forecasting in the energy sector and spent time as a guest researcher at EPFL in Lausanne. Today, he is co-founder and Head of AI at xtream, a boutique company that applies cutting-edge technology to solve complex business challenges. Emanuele is also a contract professor in AI at the Catholic University of Milan. He has published eight papers in international journals and contributed to over 30 international conferences worldwide. His engagements include AMLD Lausanne, ODSC London, WeAreDevelopers Berlin, PyData Berlin, PyData Paris, PyCon Florence, the Swiss Python Summit in Zurich, and Codemotion Milan. Emanuele has been a guest lecturer at Italian, Swiss, and Polish universities.
PolyGloT: Personalized and Gamified Learning Paths with Generative AI
This session presents PolyGloT, an innovative framework for creating and executing personalized and gamified learning paths powered by generative AI and integrated into immersive virtual environment.
The talk will explore how PolyGloT, developed at the University of L’Aquila, leverages AI models to personalize educational experiences. By embedding and fine-tuning large language models (LLMs) using educational data, PolyGloT supports the dynamic generation and evaluation of learning content tailored to each student's unique needs.
Key topics include:
The architecture of PolyGloT: combining a visual learning path editor, an execution engine, and generative AI.
Adaptive learning paths: how AI recommends and personalizes activities in real time based on learner performance.
Assessment and feedback: automated progress tracking, formative evaluation, and student feedback via AI.
Integration with WorkAdventure: delivering gamified, collaborative, and spatially-organized digital classrooms.
Attendees will gain insights into the design, implementation, and impact of a next-generation educational platform that fosters engagement, accessibility, and personalization—paving the way for scalable, AI-enhanced learning environments.
Antonio Bucchiarone is currently an Associate Professor at the University of L'Aquila. He is specialized in adaptive systems, motivational digital systems using AI and gamification and the design and development of educational tools. His research focuses on models and tools for adaptive (collective) systems, modeling and development languages for motivational digital systems, and educational tools supported by and based on Model-Driven Engineering and AI. Dr. Bucchiarone has contributed to numerous national and European R&D projects, including the Erasmus+ project ENCORE. He is a Senior Member of IEEE and a member of the Italian chapter of the IEEE Intelligent Transportation Systems Society. He also serves as an Associate Editor for several journals, including IEEE Software, Journal of Object Technology (JOT), IEEE Transactions on Education, IEEE Transactions on Games, IEEE Transactions on Intelligent Transportation Systems (T-ITS), and IEEE Technology and Society.
Automated design of Agentic Systems: a new frontier in AI architecture
The automated design of agentic systems represents a newly emerging research frontier focused on the creation of intelligent agents by other intelligent agents. This paradigm introduces a meta-agent capable of iteratively generating, evaluating, and refining agent designs, encompassing everything from prompt engineering and tool use to workflow synthesis. In this talk, we’ll explore how this recursive, generative approach enables the discovery of novel agentic behaviors and architectures, potentially unlocking general-purpose automation at scale. We will discuss current methods, challenges, and the transformative implications of teaching machines to autonomously design better versions of themselves.
ML Engineer
AI Agents for Quality Management - lessons from the field
The session will be structured as follows:
Introduction - 2 mins: We begin by exploring the use of AI agents in quality management, examining why quality management is critical—especially within regulated contexts such as Aerospace, where compliance and precision are non-negotiable. The introduction will highlight the journey from traditional, manual methods (like pen and paper) to the advanced, AI-driven solutions now transforming the quality landscape.
Identifying Challenges - 3 mins: The session will delve into the specific challenges faced in quality management, particularly in regulated industries. These include the difficulty in obtaining clear answers from vast data sets, the complexity of locating relevant information, and the significant time investment required for manual data entry. The struggle to turn existing data into actionable insights will be a focal point of this discussion.
AI Opportunities - 2 mins: Next, we’ll discuss how AI technologies present new opportunities to overcome these challenges. The focus will be on how AI can simplify and enhance the way information is accessed and used in quality management. This includes shifting from traditional, search-based workflows to more intuitive conversational AI experiences. Attendees will see how integrating various modalities—such as text, voice, and images—enables more natural and efficient interactions. The generation of dashboards and the ability to correlate documents and numerical data in real time will also be highlighted as key benefits.
The Solution: LIV4 - Quality - 2 mins We then introduce LIV4, our AI-powered solution designed to revolutionize quality management: LIV4 enables users to create, search, and interact with documents using multiple input types, including writing, speaking, and taking photos.
The platform empowers users to generate dashboards and seamlessly correlate diverse types of information, making complex quality management tasks more efficient and accessible.
Architecture and lessons learned 20'
Matteo is a startup founder and product leader passionate about the intersection of technology, business, and innovation. With a background in automation engineering and international experience in both China and the US, he has led global teams and driven the development and launch of AI and IoT solutions for quality management. Currently, he is building a generative AI startup, leveraging a blend of hands-on engineering and strategic product management expertise to transform how organizations harness the power of AI in the domain of Quality Management.
Espandere Microsoft Security Copilot con agenti AI
Questa sessione fornirà inizialmente una panoramica completa delle capacità standalone e embedded di Microsoft Security Copilot per rafforzare la sicurezza delle organizzazioni in un panorama di minacce informatiche in continua evoluzione anche grazie all’utilizzo di AI. Recentemente Microsoft ha introdotto una serie di agenti autonomi progettati per migliorare la sicurezza e gestire compiti ad alto volume, come il triage degli attacchi di phishing e la gestione delle vulnerabilità. Discuteremo quindi come questi agenti, integrati con soluzioni di sicurezza esistenti come Microsoft Defender, Microsoft Entra e Microsoft Purview, possono aiutare le organizzazioni a rispondere in modo più efficace alle minacce informatiche.
Mario Serra lavora come STrainer & IT Security Specialist in Overnet Education. Da più di 20 anni è nel nel settore dell'IT è ha così potuto approfondire la sua conoscenza in molti ambiti tra cui quello della security e infrastructure. Nel mondo Microsoft è particolarmente skillato su Active Directory, Soluzioni di Security e Compliance, soluzioni di virtualizzazione, Microsoft Azure e sui prodotti della famiglia Microsoft 365. Nel corso del tempo ha conseguito differenti certificazioni Microsoft in Identity and Access Management, Compliance, Active directory, Server Virtualization e Server Administrator. E' stato Microsoft MVP per la categoria Directory Services, Remote Desktop Services e Enterprise Mobility. E' ora MVP per la categoria Security. Ha partecipato come speaker a molte conferenze come ad esempio WPC, BeConnected Day, Azure Day, Global Azure, Aperiteam Conference parlando sempre di tecnologie Microsoft.
Laureato in Scienze dell'Informazione e MCT trainer specializzato principalmente nell'ambiente IT, Security e Formazione su differenti tecnologie Microsoft da Windows Server a SCCM in ambienti locali, da Microsoft 365 ad Azure passando per Microsoft Endpoint Manager in ambiente hybrid cloud. Lavoro come dipendente part-time (50%) presso il Dipartimento di Informatica dell'Università degli Studi di Milano, dove svolgo l'attività di sistemista Microsoft, e come consulente libero professionista nel restante tempo. Come consulente svolgo attività di progettazione, implementazione, troubleshooting di ambienti ibridi basati su tecnologie Microsoft, progettazione e implementazione di funzionalità di sicurezza delle suite Microsoft e attività di docente/speaker. Da Agosto 2020 Microsoft MVP nella categoria Entreprise Mobility mentre da Gennaio 2023 nella categoria Security; da Febbraio 2021 CTO di Overnet Education.
Agentic AI with local LLMs/SLMs: why not?
When talking about creating agentic AI solutions, everyone immediately links this to adopting a standard online AI model. But there's also an emerging world of AI models (LLMs/SLMs) that are open and that can run fully offline with zero costs for an organization. Can those models be used on creating real-world agentic AI solutions? In this session we'll walk through the world of the open AI models. We'll see what an open model is, what are the main open models, they're pros and cons and how you can create agentic AI solutions that can run fully offline using a local LLM by using frameworks like Semantic Kernel, Autogen, the new Foundry Local service and more. After this session, some new scenario of AI adoption will probably open up for you.
Stefano Demiliani is a Microsoft MVP for Business Applications and for Azure, MCT Regional Lead for Italy and a long-time expert on different Microsoft technologies. He works as a CTO for EID - Lodestar spa and he's frequently involved on helping customers on moving their business to the Azure cloud. He's the author of many successful IT books related to Dynamics NAV, Dynamics 365 Business Central, Azure DevOps, Azure solution development and Power Platform. He’s a Microsoft Community Champion for Azure. He's the winner of the global "Tech Start of the Year" Digital Revolution Award 2023 (the "Tech Star of the Year" award recognizes outstanding contributions to the cloud industry by individuals working at any level within an organization).
Agenti Intelligenti e Dati Aziendali: Innovazione con Teams, Fabric e Copilot Studio
Questa sessione approfondisce l'integrazione avanzata di Microsoft Fabric, Copilot Studio e Teams per ottimizzare la ricerca e l'analisi dei dati aziendali attraverso il linguaggio naturale. Inizieremo con Fabric, una piattaforma di gestione dei dati su larga scala, che consente di raccogliere, archiviare e analizzare enormi volumi di informazioni aziendali. Utilizzeremo i nuovi Data Agent per istruire il nostro modello alla ricerca dei dati tramite l'intelligenza artificiale.
Successivamente, esploreremo Copilot Studio, uno strumento di automazione che sfrutta l'intelligenza artificiale per semplificare i l'automazione degli agenti. Infine, useremo Microsoft Teams, che grazie all'uso del linguaggio naturale, gli utenti possono interagire con i dati in modo intuitivo, migliorando la comunicazione e la collaborazione aziendale.
Questa sessione dimostrerà come l'integrazione di Fabric, Copilot Studio e Teams, supportata dal linguaggio naturale, possa trasformare la gestione dei dati aziendali, rendendo le operazioni più efficienti e le decisioni più strategiche.
(CTO) at Lobra Futura Core competencies: Azure: I have extensive experience in using Microsoft Azure to create scalable and reliable solutions Power Platform: I am an expert on the Power Platform where I have designed Enterprise architectures and supported several major customers Data Platform and AI: I am passionate about data-driven solutions. Implement real data-driven services particularly with Power BI and Microsoft Fabric. Also has experience in artificial intelligence and data analytics.
Explore AgentCon
AI Agents Developer Conference
AgentCon made possible by:
