Search posts, tags, users, and pages
alexeyg377
ML Platform & Distributed Systems Engineer
Agentic systems inherit every classic distributed systems failure mode plus one they don't have a name for yet: an agent can fail completely and still sound like it succeeded. A look at which reliabil
Andrew hi,
Thank you for feedback Probably it depends on the use case, but I suppose a validation layer could be more than another LLM call
Andrew
The distributed-systems toolkit transfers more than expected, useful, but the validation layer is just another LLM call that can also be confidently wrong. That's the part the vocabulary doesn't cover
alexeyg377
ML Platform & Distributed Systems Engineer
Andrew hi,
Thank you for feedback Probably it depends on the use case, but I suppose a validation layer could be more than another LLM call