As we move past the era of simple chatbots, the challenge of AI shifts toward scalability and interactivity. This event brings together two critical pillars of modern AI development: robust infrastructure and adaptive user experiences.

Join us for two deep-dive sessions where we explore how to build agentic workflows that scale to thousands of instances on Google Kubernetes Engine (GKE) and how to manifest those agents through A2UI, Google’s open-source framework for "living" Angular interfaces.

Agenda:

02:00 PM | A2UI: Building "Living" Interfaces

Speaker: Ezéchiel Amen AGBLA

The Deep Dive: Learn how to build software ecosystems where the interface adapts instantly to user intent and task context.

The emergence of autonomous agents based on large language models (LLMs) is radically transforming human-machine interaction. However, the transition from a simple text-based chatbot to a dynamic, actionable graphical interface remains a major technical challenge. A2UI is an open-source project designed to bridge this gap by providing a standardized architecture that enables AI agents to manipulate, generate, and react to user interface (UI) components in real time.

Unlike traditional static approaches, A2UI enables a symbiosis where AI does not merely display data, but becomes the architect of the user experience, adapting views based on the user’s intent and the task context. This project aims to provide developers with the tools needed to build software ecosystems where the interface is “alive,” thereby facilitating the deployment of complex and intuitive applications driven by artificial intelligence.

03:30 PM |Beyond the Prompt: Scaling Kagent Workflows on GKE

Speaker: Ashutosh Bhakare

The Deep Dive: As AI agents move from local experiments to enterprise requirements, the underlying infrastructure becomes the bottleneck. This session explores how to deploy Kagent on Google Kubernetes Engine (GKE) to create a self-healing, auto-scaling agentic ecosystem.

We will dive into:

• Architecting for Autonomy: Utilizing GKE’s Autopilot and GPU slicing to optimize Kagent’s compute-heavy tasks.

• State Management: Implementing persistent memory for agents using GKE-native storage solutions.

• Observability: Using Cloud Monitoring to track agent "reasoning" latency and resource consumption.

Attendees will leave with a blueprint for a production-ready agent framework that scales from one agent to thousands without breaking the platform.

04:30 PM | Closing Remarks