沒什麼組織、一些倏忽即逝有趣或不有趣的想法
From Core Vision to MVP through LLM
Windsurf seems to provide an excellent way to manipulate agents in IDE
PR, Issue, Merge, through gh CLI
New Use Mindset
LLM observations
API Rate Limits seems to be a problem when we are heavily using LLM tools, so people tend to use multiple LLM based tools and switch between them.
Business isn’t really caring about how the product or code is generated, as long as the product is useful, eyes catching enough then it’s a good way to make it. However, it might not be a good way to run it long term. The needs change everyday so the product should change rapidly according to the user feedback. We still need to figure out how to interact with LLM the right way to achieve it.
The LLM tools seems to forget the mobile environment, many tools are not supporting the Xcode, Android Studio, etc. XD
As freelancers or engineering consultants products are mostly one-off, which would rely heavily on chat to complete coding tasks without touching the code.
Therefore, Windsurf's cascade flow enables them to cooperate seamlessly with AI, as AI in chat knows user’s edits and can continue what user has done in chats.
Team leads found it is difficult to introduce Windsurf / Cursor / Aider chat capability to their team members. Team members often do not know how to properly instruct the AI agent to complete tasks.
Found New Tools - bolt.new, Google Studio AI, Windsurf