Beyond the Model: Infrastructure for Reliable Agentic Workflows

Do not forget to register here to secure you spot: https://www.eventbrite.it/e/beyond-the-model-infrastructure-for-reliable-ai-agents-in-specialised-doma-tickets-2004513544973

AI agents are changing how people discover products and how teams get work done. Across three talks, we’ll explore how brands can reach buyers through AI assistants, how agents can work alongside people, and how organizations can keep the information their agents rely on consistent and up to date.

Agenda
18:30 - Doors open
19:00 - Talks begin
20:30 - Networking aperitivo (courtesy of Wonderful)

Talk 1
Title: Agents Don’t Work Alone.
Abstract: The industry talks about agents as autonomous digital workers, but no useful worker operates in isolation. Agents depend on colleagues, tools, information, processes and supervision. This talk translates that organizational reality into software architecture, showing how an AI OS combines multi-agent collaboration, workflow engines, enterprise systems, data, user interfaces and human control.
Bio: Saverio Murgia has been working in AI since 2014, when he co-founded Horus Technology to improve the lives of nearly 300 million people who are blind or visually impaired, leading the company to a successful exit in 2019.
A Forbes 30 Under 30 honoree, he then led product development at Caricami and served as CTO at a Los Angeles-based startup studio, helping launch over 20 ventures across SaaS, AI, and B2B/B2C products.
Today he is CTO Italy at Wonderful, the $2B enterprise agent platform for designing, deploying, governing, and scaling AI agents across complex workflows, markets, and regulatory environments.

Talk 2
Title: Why Your AI Agents Disagree: Fixing Context Drift Across Your Organisation

Abstract: Ask two colleagues’ AI agents the same question, and they may get two different answers. Each agent relies on its own mix of prompts, skills and reference material, often copied, rarely reviewed and increasingly out of date. This session explores why context drifts across an organisation and how to manage it like data.
Bio: Gianmario Spacagna is an AI engineering leader and the founder and CEO of Coelevate, a Rome-based startup building a Company Brain platform powered by proprietary continual learning technology. He has more than 14 years of experience turning ambitious technology visions into production products, combining hands-on engineering with strategic direction. His work covers agentic and LLM-based architectures and large-scale machine learning pipelines serving hundreds of millions of users. He focuses on agentic AI, applied machine learning and intelligent product design.

Talk 3
Title: Buying Went Agentic. Selling Is Next.
Abstract: Today, buyers increasingly rely on AI assistants to find what they need, while many brands still sell through static pages built for a person to browse. In this talk we introduce Veliu, the agentic commerce lab working on how brands can be understood and chosen by AI agents without losing their voice or their customers’ trust. We start from the Agentic Selling Manifesto to explain why this shift affects both those who search for products and those who sell them, and we show how discovering content and products is becoming a conversation. We close with the practical implications for brands, publishers and consumers, and with the open questions around transparency and control.
Bio: Gianmaria Monteleone is founder and CEO of Veliu, the agentic commerce lab building technology for businesses and consumers. He co-founded Veliu in 2024 with Alessandro Ruffini and regularly shares his views on how brands should prepare for buyers who shop through AI assistants. Before Veliu, he founded SENSO, a Milan-based company that refurbishes and sells iPhones using robotics and proprietary software, after launching a franchise chain of technology stores in his early twenties. He also mentors startups at Techstars and Founder Institute.