At PEC 2026, my main contribution was helping other people share their work. On September 12 in Beijing, our LangGPT community organized the Harness Workshop track, bringing together practitioners to exchange recent approaches and application experience. I often attend conferences as a guest speaker. This time, I took on more of an organizer’s role: helping create the conditions for speakers to present well and for participants to have useful conversations. Our team handled the workshop from preparation through delivery. That included inviting speakers, shaping the topics and session flow, preparing presentation templates and individual speaker posters, and writing introductions and transitions. On the day, we coordinated the schedule, managed presentation materials and screen switching, and supported the different requirements of talks and live demonstrations. The practical details mattered: who needed their slides on the presentation computer, who needed a separate connection for a demo, and how to move smoothly between sessions. The workshop focused on AI working methods built around real tasks. Speakers discussed how to move from context to execution and delivery, including defining the task, using tools, producing an artifact and providing evidence that the work had been completed. That practical focus also shaped how we organized the event. We wanted to give people enough support to concentrate on their ideas and implementations, and create room for direct exchange. Organizing gave me a different view of community work. You pay attention to the whole session: how the contributions connect, whether the setup supports each speaker, and what the audience can take away. I enjoyed serving the community in this way. Bringing practitioners together and supporting their sharing is a contribution I value alongside giving my own talks. Thank you to the PEC organizers, every workshop speaker, the LangGPT team and everyone who joined the discussion. It was a pleasure to help make this exchange happen. #LangGPT #PEC2026 #AIHarness #AIAgents #DeveloperCommunity

Personal Agent 的终点不该是一个更聪明的命令行，人天然习惯使用有直观视觉反馈的工具。Eazo 的这波设计理念确实很有意思，分享几个实用场景~

非常顶的开源项目，一键帮你自动剪辑，实现爆款视频复刻，测了很多case，感觉还是很不错的，具体看评论区了！

AI 厂商的「高低峰价格」策略终于影响到了上班时间.....