A single model optimized the parser. The real bottleneck was governance(council.expressible.ai)
2 pointsby veniyer7 hours ago2 comments
  • veniyer7 hours ago
    I’ve been experimenting with multi-perspective reasoning for architecture reviews, and this SIEM case was one of the clearer examples where the bottleneck wasn’t where we expected. Curious if others have seen similar cases where tooling gets optimized but process is the real constraint.
  • elee2517 hours ago
    this is the type of discussions we have on slack and zoom meetings. Would be great to collaborate here
    • veniyer6 hours ago
      That's exactly what we noticed — these discussions happen across Slack threads, Zoom calls, and Confluence pages, and the reasoning gets lost. The approach we tried was structuring that deliberation so the tradeoffs and confidence shifts are tracked as the conversation evolves. Happy to share more about how we set it up if you're interested.