Tackling the hardest questions in agent development with Philipp | The Agent Factory Podcast

Опубликовано 26 августа 2025, 23:38

Ready to listen about how to use Gemini for building applications like Deepmind Deep Research agent? In this special episode, we're joined by Philipp Schmid from Google DeepMind, a leading voice in the AI community, to answer some of the toughest, most practical questions developers face today. We skip the agent industry pulse and dedicate the entire episode to a deep-dive masterclass on agent architecture, development, and evaluation.

In this episode you will learn:
1️⃣ Model & framework selection: A pro's take on choosing between Gemini and open models like Gemma, plus a breakdown of when to use frameworks like LangGraph, CrewAI, or the ADK.
2️⃣ The agent development lifecycle: A behind the scenes look at how the "Deep Research Agent" was built, from initial concept and prompt engineering to multi-agent patterns.
3️⃣ Gemini CLI pro tips: Discover how to leverage the Gemini CLI for advanced agentic workflows and see a demo of its powerful capabilities.
4️⃣ Model agnostic prompts: Philipp answers a tough community question about whether it's possible to write prompts that work across different model families (like Gemini, GPT, and Claude) without a major loss in performance.

🚀 ANNOUNCEMENT: Accelerate AI with Cloud Run Take your agents from prototype to production! Join our new event series to learn how to deploy scalable, secure, and production-ready agents. Register for the North America series now, with more global locations coming soon! Register Here → goo.gle/3JAflqs

Join the conversation on social media with the hashtag #TheAgentFactory.

About The Agent Factory:
"The Agent Factory" is a video-first technical podcast for developers, by developers, focused on building production-ready AI agents. We'll explore how to design, build, deploy, and manage agents that bring real value using the latest open-source tools and frameworks. Going beyond basic AI/ML concepts, the podcast will delve into the practical aspects of agentic AI, covering everything from single-purpose agents for task automation to complex multi-agent systems.

Connect with Us & Ask Questions:
- Leave your questions and feedback in the YouTube comments below!
- Connect with the community at the Google Developer Program forums → goo.gle/4mUGqD2

🔗 Resouces & links mentioned:
➖ Follow Philipp Schmid on X → twitter.com/_philschmid
➖ Read Philipp's Technical Blog → philschmid.de
➖ Learn more about the Deep Research Agent → goo.gle/4mzVJkV
➖ Gemma 3 270M - The compact model for hyper-efficient AI → goo.gle/4lSdMll
➖ Effective Prompt engineering → goo.gle/3HGnjOn
➖ Philipp's blog post on Context Engineering → goo.gle/4nmGdJp
➖ Philipp's blog post on Gemini CLI → goo.gle/3HGoFbV
➖ Philip’s demo showing how to use Code Sandbox MCP with Gemini CLI → goo.gle/4n8qL3r
➖ Gemini CLI Documentation → goo.gle/3JxDRZq
➖ Gemini CLI custom slash commands → goo.gle/4685TmV

#AIAgents #Gemini #DeepMind #TheAgentFactory

🔔 Subscribe to Google Cloud Tech → goo.gle/GoogleCloudTech

Speaker: Ivan Nardini, Philipp Schmid, Shir Meir Lador
Products Mentioned: Gemini, Langraph, Gemma, Gemini CLI