Docky

Docky is 张月光’s current agent product direction in 130. 张月光创业两年首次访谈：妙鸭不是AI Native产品、流程到上下文设计、One Way Door和乙女游戏. The first wedge is PPT generation, chosen because the category has product-market fit but existing AI PPT tools have not become One Way Door Product experiences.

The source frames Docky as an AI Friend Products attempt rather than a pure automation tool. Zhang wants an agent that helps users cross ability boundaries, with practical value and emotional attachment gradually blending when the same object repeatedly helps solve hard tasks. His product approach is to refine the smallest generated unit and needed context first, then reverse-engineer the user flow.

Product Claims

• PPT is a useful entry because users already need presentations, but templates, code-generated slides, and image-generated slides each create serious editability or quality limits.
• The intended agent value is “ability breakthrough,” not only end-to-end task outsourcing.
• The agent loop may be short, frequent, and low-latency rather than only long-running offline delegation.
• The product should preserve human participation by expanding user capability instead of simply removing the user from production.

Connections

• 张月光 — founder committing company resources to Docky.
• AI Friend Products — broader AI-friend frame behind the product.
• AI Native Product Design and Context Engineering — method of starting from model context and smallest useful generated units.
• Agentic Workflow and Human-Agent Collaboration — workflow pattern Docky belongs to.
• One Way Door Product and Product Led Willingness To Pay — standard Docky must meet to matter commercially.