https://video.ut0pia.org/videos/watch/38c6282c-9fbb-4948-870d-0a2e862a7aec AI agents that book 15 guests in a 10-person room. Agents that fabricate statistics when data doesn't exist. Agents that pick wrong tools from 29 options, wasting $47 in tokens. These aren't prompt engineering failures, they're architectural limitations that need structural solutions. This hands-on workshop covers 5 research-backed techniques to prevent agent hallucinations: Graph-RAG (Neo4j) - Replace vector similarity guessing with precise entity relationships. Result: 73% fewer fabricated statistics., Semantic Tool Selection - Filter 29 tools to the relevant 5 using embeddings. Result: 89% token reduction, accurate tool selection., Multi-Agent Validation - Executor-Validator-Critic swarms catch fabrications through cross-checking. Result: 92% detection rate., Neurosymbolic Guardrails - Framework-enforced rules (lifecycle hooks) that agents cannot bypass. Result: Zero business rule violations., Agent Steering - Guide agents to self-correct instead of blocking them. Result: Task completion without hard failures., Each demo includes live code, before/after metrics, and failure case analysis. Final module shows production deployment. You'll walk away with working Python implementations, a decision framework for when to apply each technique, and an open-source repository adaptable to your domain. code: https://github.com/elizabethfuentes12/why-agents-fail-sample-for-amazon-agentcore Speakers: Elizabeth Fuentes (AWS): Elizabeth Fuentes is a developer advocate and AI engineer focused on what makes agents fast, cheap, and correct in production. She turns failure modes (hallucination, token blowups, context overflow, lost memory) into named, measurable fixes, each backed by a runnable demo and before/after numbers. Her work covers the architectural decisions behind reliable agents: context offloading, the split between conversation and data memory, semantic versus exact-reference retrieval, guardrails, and agent evaluation. With 107+ published technical articles and a Master's in Data Science, she shares production agent patterns across English and Spanish developer communities, and likes turning complex concepts into something anyone can learn. X/Twitter: https://x.com/ElizabethFue12 LinkedIn: https://www.linkedin.com/in/lizfue/ GitHub: https://github.com/elizabethfuentes12 Sun, 12 Jul 2026 03:47:16 GMT https://validator.w3.org/feed/docs/rss2.html PeerTube - https://peerate.fr https://peerate.fr/lazy-static/avatars/d7b83a50-fe89-4e9f-8234-2a5bb1c39ab5.jpg https://video.ut0pia.org/videos/watch/38c6282c-9fbb-4948-870d-0a2e862a7aec All rights reserved, unless otherwise specified in the terms specified at https://peerate.fr/about and potential licenses granted by each content's rightholder.