Agentic RAG Platform for AI-Assisted Workflows
summary
An end-to-end Agentic RAG platform that ingests PDF, PowerPoint, audio and video, OCRs and chunks text, runs NER to build a knowledge graph (17 entity types), and generates embeddings from Progress or third‑party models (OpenAI, Google, Hugging Face). It offers up to 30 retrieval strategies, modular selection of embedding + LLM, six enrichment agents (labeler, generator, graph extractor, Q&A, content safety, LLM security), batch extraction across 100,000 contracts, and quality scoring via Remy LLM-as-judge measuring context relevance, answer relevance and groundness.
AI-written summary · about
chapters (16)
- 0:00Introduction and speaker intro
- 2:18What are AI agents?
- 3:40Introducing Agentic RAG
- 5:29Progress Agentic Rack explained
- 7:13Data ingestion pipeline
- 10:01Embeddings and modularity
- 12:07Retrieval strategies
- 13:50Output quality and evaluation
- 15:58Platform dashboard demo
- 18:05Using AI agents for enrichment
- 21:23Document processing example
- 25:19Search and query demonstration
- 27:05Fine-tuning retrieval options
- 30:41RackLab for model comparison
- 33:30Quality metrics with Remy
- 36:25Q&A and closing remarks
related talks
18mLLMs and GenAI: A Guide to Building Your Own RAG Model for Document Interaction
01h 16mForget RAG Pipelines—Build Production Ready Agents in 15 Mins
29mAgentic GraphRAG: Multi-Agent Knowledge Graph Construction for Research Teams
22mOrchestrating Multimodal RAG With Agentic Workflows With OpenSearch
26mAdvanced Document Parsing for AI Workflows
51mDeveloping a Custom RAG System
31mOrchestrating Multimodal RAG With Agentic Workflows
41mFrom RAGs to riches: Build an AI document interrogation app in 30 mins