Jev 把 AI 输出从文本变成可执行决策

Vikas gupta @vicky_grok原推发布于 2026.09.23

在 X 上打开原推

JEV FLIPS THE USUAL AI WORKFLOW. Most AI systems are built to generate something you can read. Jev is interesting because the output can be something your software can act on directly. Instead of: Question → Paragraph → Parse → Validate → Retry You can think: Question → Decision → Probability → Action That opens up useful patterns for: → Classification → Routing → Verification → Scoring → Agent workflows Sometimes the best AI output isn't an answer. It's a decision your code can use.

这条是引用着另一条写的：x.com/vicky_grok/status/2101988006446203119