Beyond LLMs: Building an AI-Powered Decision System at Scale
summary
Generative AI sees over 50% of projects fail to reach production and roughly 30% of failures trace to bad data the presenter shows an agentic design that stitches LLMs with large retrieval models and Large Action Models to reduce hallucinations by modular checks he shows a real example where a chain had to retry four times before a simple SQL intent was understood and demonstrates autonomous actions like flagging compliance and updating databases using frameworks such as ReAct LangChain Autogen
AI-written summary · about
chapters (6)
- 0:00Introduction to generative AI
- 3:54Challenges in productionizing AI
- 13:24Building scalable AI systems
- 19:44Agentic AI core components
- 26:43Use case and frameworks overview
- 30:35Future directions and Q&A
related talks
40mPannel Discussion | From Hype To Value: Real World Applications Of Generative AI And LLMs
29mGenerative-AI: Usecase-Specific Evaluation of LLM-powered Applications
43mAI Development Lifecycle: Learnings of What Changed with LLMs
01h 39mMastering LLMs: From Prompt Engineering to Agentic AI
46mEnhancing Agent Reliability Through Evaluation-Driven Development
44mProduction-Grade LLMs & AI Agents
41mBringing stories to life with AI, data streaming and generative agents
40mLLMs & Data Warehouses: Reliable AI Agents for Business (Keynote)