Toss Systematizes Knowledge Management with AI-Integrated Internal Documentation Platform 'todoc'

Toss's Knowledge System Team launched the internal documentation platform todoc, expanding the role of Technical Writers (TWs) from simple document authoring to product system design. todoc was developed to address the high usage barriers of existing SSG-based documentation tools and the issue of fragmented 'junk knowledge'. Within six months of launch, it recorded over 500 documents, more than 40,000 valid pages, and over 1,000 MAU.

AI Integration and Single Source of Truth

The core of todoc is building an organizational Single Source of Truth (SSoT) by easily integrating high-quality knowledge with AI. Teams can freely access todoc knowledge via Bots, APIs, CLIs, MCP, and more, utilizing it in various workflows such as running request-handling bots or organizing product specifications. Unlike existing tools, todoc lowers barriers to allow non-developers to write easily, and connects various sources like internal messengers to convert them into complete documents.

Automated Knowledge Validity Management

The true challenge of documentation is not information collection but determining 'knowledge validity'. todoc embeds the 'good document' criteria defined by TWs into the system, providing AI-based editing and automatic draft generation features. It is also designed to monitor decisions in internal workflows and code changes to ensure documents remain up-to-date. This allows TWs to focus on system design that consistently produces good documents, rather than manual reviews.

Future Vision

Toss aims for todoc to become a fully automated system that detects outdated documents, notifies owners, and suggests improvements. The ultimate goal is to automatically capture and organize team outputs without manual updates to project management tools, thereby reducing repetitive questions from new team members and increasing overall organizational work speed.