Huang Bote

Huang Bote is the founder of AirJelly and the main guest in OpenClaw 之后，谁将定义主动式 AI 的新战场？｜对谈 AirJelly 黄柏特. The source identifies him as a 24-year-old Xidian University graduate who previously worked at ByteDance as product manager for the open-source Mycontext project.

His product thesis is that the next useful personal agent needs Intent Context, OS-Level Context, Persistent Agent Memory, and execution ability. He argues that full recording creates too much low-value noise, while intent-linked events and entities can become durable memory for Proactive Agents.

Position In The Episode

• Treats Enter as the key intent boundary AirJelly wants to capture, analogous to Cursor redefining Tab for AI coding.
• Uses Open Claw and Claude Code as useful shocks: they exposed how fast execution tooling could be copied, pushing AirJelly toward context and memory.
• Argues that if Vibe Coding can build a product to 60 or 80 points, it may not be a strong startup target; AirJelly’s hard part is the context-understanding layer.
• Frames memory as the main moat for consumer agents, because accumulated user context becomes hard to transfer after weeks or months.
• Presents privacy as both product risk and adoption filter: early users may trade privacy for efficiency, but the product must still design local storage, encryption, and PII handling carefully.
• Connects debate training to startup thinking: a founder must identify the people, moments, and scenarios where an idea becomes true rather than claiming universal validity.

Connections

• AirJelly and Mycontext — current company/product and predecessor project.
• Proactive Agents, Intent Context, OS-Level Context, and Persistent Agent Memory — concepts Huang uses to define the product.
• Open Claw, Claude Code, Cursor, Manus, and ChatGPT — comparison products in the discussion.
• AI Organization Design — team operating philosophy and AI-native company experiments.