YouTube Video
Are you tired of writing endless system prompts and praying your AI agents behave? Stop prompting and start routing!
In this video, Google Solutions Engineer Miguel Gutierrez joins Martin Omander to show you how to build reliable multi-agent apps. Instead of trying to prompt your way to agent coordination, you'll learn how to use the native graph-powered workflow primitives in Google's Agent Development Kit (ADK 2.0).
We walk step-by-step through a real-world demo: building a "smart parser" multi-agent application. You'll see how we take completely unstructured, chaotic data, like forwarded emails, messy Slack transcripts, and raw PagerDuty logs, and clean it up using deterministic multi-agent workflows.
If you have a problem that requires reasoning, multi-source coordination, and self-correction, ADK workflows are the solution.
Resources:
Smart Parser GitHub Repository → https://g.dev/cloud/adk-smart-parser
Deep Dive, Deterministic Regex vs LLM Quality Gates) → https://g.dev/cloud/adk-smart-parser-key-highlights
Workflows in Google ADK 2.0 → https://g.dev/cloud/adk-workflows
Do you use agent workflows in your projects? Have you hit token limits or struggled with agent coordination? Let us know in the comments below!
Chapters:
0:00 Intro
1:20 Demo
3:44 Code review
7:06 Questions and Answers
8:29 Takeaways
9:21 Conclusion
Watch more Serverless Expeditions → https://goo.gle/ServerlessExpeditions
🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech
#AIAgents #GoogleADK #AgentDevelopmentKit #AIWorkflows #SoftwareEngineering #GenerativeAI #PythonCoding
In this video, Google Solutions Engineer Miguel Gutierrez joins Martin Omander to show you how to build reliable multi-agent apps. Instead of trying to prompt your way to agent coordination, you'll learn how to use the native graph-powered workflow primitives in Google's Agent Development Kit (ADK 2.0).
We walk step-by-step through a real-world demo: building a "smart parser" multi-agent application. You'll see how we take completely unstructured, chaotic data, like forwarded emails, messy Slack transcripts, and raw PagerDuty logs, and clean it up using deterministic multi-agent workflows.
If you have a problem that requires reasoning, multi-source coordination, and self-correction, ADK workflows are the solution.
Resources:
Smart Parser GitHub Repository → https://g.dev/cloud/adk-smart-parser
Deep Dive, Deterministic Regex vs LLM Quality Gates) → https://g.dev/cloud/adk-smart-parser-key-highlights
Workflows in Google ADK 2.0 → https://g.dev/cloud/adk-workflows
Do you use agent workflows in your projects? Have you hit token limits or struggled with agent coordination? Let us know in the comments below!
Chapters:
0:00 Intro
1:20 Demo
3:44 Code review
7:06 Questions and Answers
8:29 Takeaways
9:21 Conclusion
Watch more Serverless Expeditions → https://goo.gle/ServerlessExpeditions
🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech
#AIAgents #GoogleADK #AgentDevelopmentKit #AIWorkflows #SoftwareEngineering #GenerativeAI #PythonCoding
SOCIAL SHARE CARD GENERATOR