Knowledge Plane

Knowledge Plane is an AI-based shared memory tool designed for engineering teams. It consolidates a team's code, documentation, and chats into a unified, updated source of information to ensure consistent quality from AI tools. It bridges the infrastructural gap of fragmented memory among different AI tools, transforming amnesic chatbots into agents with long-term memory.

The functionality of Knowledge Plane is based on automatically scanning documents and codes, extracting raw facts, cleaning data, and creating knowledge cards. It combines Graph capabilities to understand dependencies with vector embeddings to locate keywords. AI agents connect to Knowledge Plane via the Model Context Protocol (MCP), eliminating the need for custom integrations. It also features workspace isolation, scoped API keys, and maintains audit trails for all queries and updates for added security.

The purpose of Knowledge Plane's auto-ingestion feature is to automatically scan documents and codebases, extract raw facts and clean data, and create knowledge cards. This functionality enables an efficient and automatic consolidation and organization of information, obviating manual effort.

Knowledge Plane's Graph capabilities and vector embeddings facilitate a profound understanding of dependencies, not just keywords. Graph capabilities model relationships between elements, while vector embeddings assist with semantic understanding. Thereby, it enables AI agents to reason about dependencies, ownership, and timelines instead of simply keyword matching.

The Model Context Protocol (MCP) is a standard protocol that AI agents like Claude and Cursor can leverage to query Knowledge Plane. It is designed to eliminate the need for custom integrations.