Description
You’ve seen AI agents work perfectly in demos, but most projects stall when transitioning from a prototype to a real-world production environment. In this webinar, Data Scientist and ML Engineer Thiago Grabe breaks down the gap between successful proof-of-concepts (POCs) and scalable, production-ready Agentic AI systems.
If you are working with LLMs or experimenting with AI agents, this session will help you understand how these systems truly behave outside of isolated demos.
In this webinar, you will learn:
- The POC Trap: Why skipping guardrails, cost controls, observability, and human escalation paths during the demo phase leads to production failure.
- Architecture Patterns: Explore the four key frameworks that actually ship, ranging from deterministic workflows with AI nodes to complex multi-agent systems.
- Failure Modes: How to identify and avoid common production killers like infinite loops, latency spikes, and cascading failures.
- Organizational Readiness: Discover why architecture only gets you 50% of the way there, and why defining roles like the "Agent Product Owner" is crucial for bridging the gap between technical and business teams.
🚀 Take Your AI Skills to the Next Level
Agentic AI Nanodegree Program
Go beyond single chatbots to engineer sophisticated, coordinated teams of AI agents.
Start learning for free today with our new 7-day free trial, which includes:
- Full access to all 50+ lessons
- The ability to build and submit 4 portfolio-ready projects
- Personalized feedback from industry mentors
➡️ https://www.udacity.com/course/agentic-ai--nd900?utm_source=organicsocial&utm_medium=youtube&utm_campaign=campaigngtm&utm_content=agentictrial
⏳ Webinar Chapters
00:00 - Introduction & Speaker Background
05:23 - The Agentic AI Landscape & Governance Bottlenecks
11:25 - The "POC Trap": Why Demos Fail in Production
24:06 - 4 Architecture Patterns for Agentic AI
28:34 - 6 Common Production Failure Modes
38:18 - Governance: Who Owns the AI Agent?
48:24 - A Suggested Roadmap for Agentic AI Deployment
53:05 - Audience Q&A
#agenticai #artificialintelligence #machinelearning #llm #aiagents #productionai #techwebinar #datascience #udacity #generativeai