Show HN: 通过 LLM 驱动安卓应用。 征求反馈
1 分•作者: philippb•7 个月前
视频：<a href="https://www.youtube.com/watch?v=OT1k7o89ed4" rel="nofollow">https://www.youtube.com/watch?v=OT1k7o89ed4</a>
我很好奇是否能通过 LLM（大型语言模型）来驱动一个应用程序。我认为这很有趣，并且在闭合智能体循环方面有一些潜在的应用场景。
请分享一下你脑海中浮现的想法！
<a href="https://www.youtube.com/watch?v=OT1k7o89ed4" rel="nofollow">https://www.youtube.com/watch?v=OT1k7o89ed4</a>
这是一个我构建完成后的原始演示。
我主要想听听你们的看法，如果这能让你们兴奋起来，有哪些应用场景？
它能变成什么，又能为你解决什么问题？
我一直在从智能体验证的角度思考这个问题，即智能体可以独立地验证错误并修复它们。
查看原文
Video: <a href="https://www.youtube.com/watch?v=OT1k7o89ed4" rel="nofollow">https://www.youtube.com/watch?v=OT1k7o89ed4</a><p>I was curious if I could drive an app via LLM. I think this is interesting and has some potential use cases to close the agent loop.
Please share your thoughts that come to mind!<p><a href="https://www.youtube.com/watch?v=OT1k7o89ed4" rel="nofollow">https://www.youtube.com/watch?v=OT1k7o89ed4</a><p>This is a raw demo after I finished building it.
I'm mostly interested in hearing from you if there are use cases that this gets you excited about.<p>What could that become and what problems would that solve for you.<p>I've been thinking about this from an agent verification point, where agents can verify bugs and fix them independently.