About Us
Redefining content production with AI technology
Background
Doc Agents originated from the 'Article Author Agent Training & Distillation System' — a complete system that iteratively trains writing Agents through reference article analysis, prompt distillation, generation verification, and human feedback loops. After multiple phases of iteration, we have validated the full pipeline from style distillation to knowledge-enhanced content production.
Technical Architecture
The system consists of three core modules: the Knowledge Base Foundation handles material storage and multi-channel retrieval (FTS5 full-text + vector semantics + tag matching); the RAG Material Enhancement pipeline automatically injects relevant materials before content generation, solving the factuality gap; the Agent Runtime provides a complete content production entry point with multi-step execution, multi-turn dialogue, and server-side persistence.
Our Values
Quality First
We don't chase generation speed — we ensure every output meets publishable standards. Factually accurate, stylistically consistent, materially rich.
Traceable & Iterable
All intermediate artifacts are versioned. Any stage can be traced, compared, and reproduced.
Open & Portable
Agent definitions are self-contained JSON files. No platform lock-in — load them in any system with zero conversion.