LLMs, Agents and Vectors: How to Build a Corporate Brain
summary
Describes building a corporate “memory” by coordinating multiple agents with LLMs to extract relevant information from unstructured sources (conversations, emails) and store it in a vector database as the system’s memory. Covers the architecture and process for agent coordination, information selection, and storage using vector embeddings; demonstrated by developers from IG Networks with AI and software-architecture experience in telecom automation.
AI-written summary · about
chapters (8)
- 0:00Project introduction
- 1:13Data sources
- 4:45Agent workflow
- 10:10Vector databases
- 15:10Tech stack
- 17:41Code implementation
- 21:34Chatbot demo
- 25:20Summary
related talks
52mAnatomy of an LLM
36mLLMs and AI agents demystified
28mEnhancing LLMs with Graph Technology
22mGenerative AI application stack and providing long term memory to LLMs
22mAgentic AI for Engineering Leaders | Strategy & Execution Guide
07mLightning Talk: Grounding LLM to Avoid Hallucinations Using Vector S...
27mMemory Engineering: Going Beyond Context Engineering
50mIntelligent Agents, the New Frontier of LLMs