Antonio Gullì is a technology leader, executive, and expert in AI and search engines, currently serving as Engineering Director for the Office of the CTO at Google (EMEA).

With a career spanning decades of innovation, he has held key leadership roles at some of the world’s most prominent tech companies. Before his current role in London, he served as Site Lead & Engineering Director for Google in Warsaw, focusing on Google Cloud. Previously, he was Vice President of Engineering at Elsevier, Principal Engineer at Microsoft (Bing), and CTO at Ask.com (Multimedia & News Search). Early in his career, he was a web pioneer in Italy, creating Arianna, the country's first search engine.

Antonio holds a PhD in Computer Science from the University of Pisa. He is also a prolific author and educator; he has written several successful technical books, including "Deep Learning with Keras", "Google Anthos in Action", and the recent "Agentic Design Patterns".

As AI evolves from reactive, chat-based interfaces to autonomous entities, the challenge for developers has shifted from writing better prompts to designing better systems. Building a truly intelligent agent requires more than just a powerful LLM; it demands a structured architecture that can perceive, reason, and act within complex environments.

In these sessions, we move beyond the hype to explore the practical Design Patterns essential for constructing robust agentic systems. Much like traditional software design patterns, these building blocks—ranging from Prompt Chaining and Tool Use to advanced Multi-Agent Collaboration and Self-Correction—provide a reusable, battle-tested language for AI development. We will dive into hands-on implementations using the industry’s leading frameworks: LangGraph, CrewAI, and the Google Agent Developer Kit (ADK). Attendees will leave with a clear roadmap for moving from simple scripts to sophisticated, maintainable, and autonomous AI systems that can solve real-world problems.

Beyond practical implementation, we will address the current frontiers of the field by deep-diving into the open research problems that still challenge even the most advanced systems. While design patterns provide a framework for today’s development, we will explore the "bottlenecks of tomorrow"—specifically the struggle with long-horizon reasoning, where agents must maintain a coherent plan over hundreds of sequential steps without losing context or diverging into "action hallucinations." We will also examine the complexities of reliable multi-agent coordination, analyzing why simply adding more agents can often degrade performance through communication overhead and error propagation. Finally, we will touch upon the critical need for verifiable autonomy, discussing how we can move toward "self-certifying" systems that provide guarantees of safety and reliability in dynamic, real-world environments.