Moving AI agents beyond “Hello World” to real production @googlecloudtech

Uploaded May 2026 | Updated September 2026, 2 weeks ago

In this interview from Google I/O, Cloud AI teams detail how to architect and maintain production systems beyond initial source generation. Learn how to combine local execution frameworks like Antigravity with managed Model Context Protocol (MCP) servers to securely aggregate real time telemetry, error logs, and multi-cloud streaming paths directly into unified observability contexts. Discover how cloud engineers can optimize microservice layers on Cloud Run, automate streaming ingest into BigQuery, and use automated multi-agent remediation chains to minimize downtime and elevate development workflows to the platform architecture level.

Watch more Google I/O Interviews → https://goo.gle/io-tech-chats
🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech

#GoogleIO #GoogleCloud

Speakers: Richard Seroter, Christina Lin, Denise Kwan
Products Mentioned: Antigravity, Model Context Protocol, Cloud Run, BigQuery

Moving AI agents beyond “Hello World” to real production

Advanced RAG techniques for better retrieval performance

Inside Google Antigravity 2.0: The complete developer guide | The Agent Factory

Beyond chatbots: How AI agents built a full OS, web apps, and more

Accelerate analytics and semantic search in real-time with AlloyDB for PostgreSQL

From notebook to production: Serving JAX at scale

Build a calendar app in AI Studio in 6 minutes

$Make your website agent ready with WebMCP Join Gemini Enterprise Agent Ready (GEAR) for the latest agent resources here. → https://goo.gle/3Srblgl Access the source code on GitHub → https://goo.gle/4g8TC5t AI agents are becoming major consumers of the web, but todays websites are built for humans, not machines. WebMCP changes that. In this video, Hugo Zanini, a Google Developer Expert in AI, shows how to expose typed tools directly inside web pages so AI agents can act reliably, quickly, and at a fraction of the token cost. Watch along and learn: * The agent shopping problem and why current approaches like screenshot-scraping fall short. * How WebMCP works to expose contextual tools directly through simple browser APIs. * A live demo of Happy Coffee (a React-based developer portal) showing Gemini in action searching catalogs and analyzing quality checks. * WebMCP for Local Agents, demonstrating how local workflows can leverage web UIs to accelerate development. Watch more How I built my agent → https://www.youtube.com/playlist?list=PLIivdWyY5sqLseBL8iUmVbdd_h30C4oEo 🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech Speaker: Hugo Zanini Products Mentioned: WebMCP, React, Gemini$

Kickstart Conversational Analytics agents with the Looker ChromeUX Block

This AI generates video lessons in seconds | The Agent Factory