shellc (@shellccn) header
shellc (@shellccn) profile picture

Software Engineer

Beijing
Joined February 2011
Photo by shellc (@shellccn)Photo by shellc (@shellccn)Photo by shellc (@shellccn)Photo by shellc (@shellccn)Photo by shellc (@shellccn)Photo by shellc (@shellccn)Photo by shellc (@shellccn)
In a keynote address during the 2025 @StanfordSCCEI China Conference, Sean Stein, president of @USChinaBusiness, cautioned that strategic miscalculations and trade tensions have left the U.S. economy with lasting setbacks—and few clear gains. ow.ly/U4K250W2Yll
1
1
327
Replying to @LaurenceMister
试图对人类的行为惯性进行量化。这种惯性可以被量化证明存在,却不容易建模预测。而且不同时间尺度上,有效性不同。
149
User pain points identified from over 18,000 feedbacks. Automatically generated by LLM without any manual changes. Whoever is interested, feel free to follow and star. github.com/shellc/pain-point… #buildinpublic #indehacker
1
46
User pain points identified from over 18,000 feedbacks. Automatically generated by LLM without any manual changes. Whoever is interested, feel free to follow and star. github.com/shellc/pain-point… #buildinpublic #indehacker
1
62
The hallucination of LLM is a bad thing, but when we let LLM generate creativity, hallucinations may actually help discover new ideas. Of course, human involvement is necessary in the final creation of ideas.
1
1
78
Just collected 18,000+ user feedback on SaaS & mobile apps, using LLM for analysis. It'll pinpoint pain points & needs, even generate product ideas. Need help on presenting final results, any suggestions? #buildinpublic
2
2
89
Replying to @HeySophiaHong
我正在做类似的事情,欢迎感兴趣的朋友交流。
1
786
#buildinpublic I am developing a process to collect feedback from internet users about products and analyze their pain points and potential needs. Is anyone interested?
5
4
1,335
LLM boosts AI's natural language understanding. Web automation via AI tools like iauto and Chrome's "Record and Replay" feature streamline tasks. Integrating AI with browser recording enhances web automation capabilities. shellc.cn/iauto/tutorials/br…
1
85
Replying to @shellccn @pyautogen
iauto also supports using low-code programming with Appium @appiumco and Playwright. By combining LLM and Autoge, you can build more intelligent automation workflows.
57
iauto uses autogen to build multi-agent systems. You can define powerful multi-agent systems through iauto and autogen simply by writing YAML files. You can combine RPA with autogen to build a more intelligent RPA. @pyautogen
1
53
Replying to @pmddomingos
以为找到了廉价劳动力,其实是个醉汉。为了让这个醉汉驾车从上海到北京,要为他专门修一条高速公路。接下来要么能让他清醒要么让他干点合适的事。微软在卖醉汉搭售保姆,创业公司在修高速公路,Open AI试图让他清醒。
77
Today we're sharing the next milestone in our Seamless Communication research — a new family of AI translation models that preserve expression and deliver near-real time streaming translations. More on this new work ➡️ bit.ly/3uBZAYG More on the individual models 🧵
34
335
88
1,564
729,054
🚀 Discover Myla - MY Local Assistants! Your gateway to self-hosting AI assistants compatible with OpenAI. It's your ideal partner for developing and running AI assistants in the local environment, seamlessly integrating with Xinference. github.com/muyuworks/myla
1
7
496