TechPrufer
Get started
AI Engineer
AI Engineer logoAI Engineer · Feature trail

Knowledge Management for AI Systems

3 videos · Oct 2026

What this trail covers

This post contributes to the understanding of knowledge management in AI systems by proposing a framework that distinguishes between prompt, memory, and weights as distinct tools for different jobs, rather than a linear progression. It emphasizes that architectural decisions about knowledge placement are made implicitly through product development and offers diagnostic questions to guide these decisions. The post also introduces a circulating architecture where agents improve through continuous learning and a relational context engine that integrates structured data (PRs, Slack threads, incident history) with unstructured data, significantly reducing token burn and improving agent performance on tasks requiring structural understanding beyond simple vector search.

All AI Engineer trails
  1. Beyond RAG: A Relational Context Engine That Cuts Token Burn — UnblockedThis post introduces a 'Relational Context Engine' as an advancement beyond traditional RAG. It addresses the limitations of vector search for answering structural questions about work (e.g., 'which PRs did I merge last week?'). The engine automatically discovers schemas, resolves identities across systems, validates queries to prevent dangerous operators, implements retries that feed errors back to the model, and integrates full-text search. This approach demonstrably reduces token burn (up to 50% in a demo) and improves agent accuracy by providing structured, contextual information. A simulator is also provided for testing the effectiveness of pre-gathered context.
  2. Why Your Company Needs a Context Graph (and How to Build It) — Gil Feig, MergeThis post introduces the concept of a 'context graph' as a company-wide brain for AI agents, extending existing knowledge management frameworks. It details an architecture that combines synced copies of third-party data with live API lookups, managed by a router and summarizer. The context graph explicitly includes company-specific skills and memory, emphasizing traceability and provenance of data. It proposes four tiers of context, including derived summaries, and a decision flowchart for selecting between cached and live data, addressing the limitations of MCPs alone for answering complex, historical queries.
  3. Stop Fine-Tuning to Fix Retrieval Problems — Anant SrivastavaThis post introduces a framework for managing knowledge in AI systems, differentiating the roles of prompts, memory, and weights. It argues that prompt engineering is for small, stable behaviors; memory is for current, large, citable, or access-controlled knowledge; and weights are for reflexes that have stopped changing. The talk highlights how architectural decisions about knowledge placement are often made accidentally through product work and presents a diagnostic for choosing the right tool. It also proposes a circulating architecture where agents improve by performing their jobs, suggesting a dynamic and evolving knowledge system.