Optimizing Complex Workflows with Event - Driven - Multi Agentic Approach
summary
Claims an event-driven, multi-agent pipeline that pairs Kafka/Pulsar-style streams with vector stores and LLMs to scale RAG and complex workflows; shows code (not yet running) and cites using vector DB + retrieval, Kafka/Pulsar for async routing, and libraries like LangChain4j, Autogen and CrewAI. Notes practical constraints: chunking 180-page PDFs (180-page PDFs), expensive LM calls (~ $4 per query), debugging and observability challenges, and that only exploratory prototypes exist so far rather than a production system.
AI-written summary · about
chapters (8)
- 0:00Introduction to AI and agents
- 8:27Generative AI and limitations
- 18:18Event-driven AI fundamentals
- 26:46Applying event-driven RAG
- 29:59Agentic workflows and design
- 41:58Agentic design patterns
- 55:15Challenges and future outlook
- 57:23Resources and closing
related talks
51mOptimizing Complex Workflows with Event-Driven Multi-Agentic Approach
50mHarnessing Event-Driven and Multi-Agent Architectures for Complex Workflows in Gen...
29mAgentic GraphRAG: Multi-Agent Knowledge Graph Construction for Research Teams
16mEffective agent design patterns in production
27mAI-Powered ETL Pipeline Orchestration: Multi-Agent Systems in the Era of Generative AI
01h 00mReliable Agentic Systems need Durable Execution 🤖
50mUnpacking an Agentic Workflow
50mAI Agentic Workflows & Design Patterns