Designing a Company Knowledge Base from Unstructured Communications
AvenaRobotics · reddit · 2026-09-01
A company without an ERP system plans to build a knowledge base from scratch using unstructured inputs like emails, Teams chats, and meeting transcripts. The main challenge is ensuring knowledge quality amidst conflicting, outdated, or inexpert information. The company plans to weight statements using a structured model of employee competencies and authority, tracking source, timestamp, and confidence. AI agents will use this system to monitor tasks and workflows. The author seeks advice on the optimal architecture—such as temporal knowledge graphs, event sourcing, or a relational database with a semantic layer—to maintain relevant and trustworthy context as data scales.
More from coding & agent
- Hermes Agent monitors chats and emails to auto-create tasks and draft work — intellectronica · 2026-09-01
- SparkLLM open-sources X2.5 4B/1.7B on-device models with native 1M-token context — vllm_project · 2026-09-01
- Ragas: Open-source framework for LLM evaluation and test set generation — goyalshaliniuk · 2026-09-01
- LightRAG: Simple and Fast Retrieval-Augmented Generation — goyalshaliniuk · 2026-09-01
- txtai: All-in-One AI Framework for Search and Workflows — goyalshaliniuk · 2026-09-01
- GraphRAG: Knowledge-Graph RAG for Complex Data Relationships — goyalshaliniuk · 2026-09-01