Knowledge as Code: A New Collaboration Architecture for AI-Native Teams

Talk overview

As AI agents become active participants in software development, research, and knowledge work, project documentation can no longer remain a static artifact written and maintained manually after the work is done. For an agent, project knowledge is not merely documentation—it is a runtime dependency that shapes every decision and action. This talk introduces “Knowledge as Code”: an approach that gives project knowledge the same properties we expect from source code—clear structure, version history, addressability, machine readability, and explicit governance. It will also present ProjectBrain.md, an Apache-2.0 open standard for creating a repository-local project brain through a BRAIN.md protocol entry point and a structured brain/ directory. The project brain preserves current project truth, architectural decisions, constraints, rationale, and their evolution in a format that both humans and agents can read and maintain. The talk proposes a new division of responsibility for AI-native teams: humans define intent, make judgments, approve important changes, and validate outcomes; agents continuously compile, update, connect, and inspect project knowledge. The goal is not a knowledge base without human involvement, but one without the burden of manual maintenance. Attendees will leave with a practical architecture for turning conversations and decisions into persistent project knowledge, executable agent contracts, and verifiable execution loops.