Scaling Specialist Knowledge with AI
summary
Red Hat builds a multi-tool AI agent that moves beyond static RAG to call tools via a Model Context Protocol, pulling live account signals from People.ai, pricing and quoting systems, and internal docs. Stack includes Gemini, Claude, IBM Granite, PGVector, Langraph/LangFuse, OpenShift and Postgres embeddings; ingestion hits ~600k documents. Evaluation shows >90% product‑fingerprint accuracy and near‑100% context recall on many sources; a UAT for ~1k users is active with an 8k salesperson rollout planned.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:32Sales credibility gap
- 5:52Virtual sales assistant
- 8:26AI system architecture
- 14:54Live demo: RAG foundation
- 17:29RAG limitations for sales
- 18:10Multi-tool AI agent
- 21:41Open-source agent template
- 22:24Live demo: multi-tool agent
- 25:11Evaluation frameworks
- 28:44Future benefits and plans
- 31:00Q&A
related talks
40mRolling out Enterprise AI: Tools, Insights, & Team Empowerment
21mHarnessing Technological Innovation & Accelerating Uptake
23mScaling GenAI Applications and Agentic AI for Business Competitive Differentiation
39mA New Stack for the Gen AI Era
14mEnterprise AI Transformation
30mRyoma: Building Open Source AI Agents for Enterprise Data Analysis
50mPractical Agentic RAG with SpringAI in Modern Enterprise Applications
24mScalable GenAI in Production