sheldonksalmon 3 karma 12d on HN HN profile →
I evaluate AI outputs used in real business decisions, and tell you exactly where human oversight is required. Most founders using AI don't know which outputs are safe to rely on, where hallucinations are creating hidden risk, or what their actual exposure looks like.

I make that clear in a structured written report, in plain language. I use a structured multi-axis evaluation method I've developed over several years. The report shows you the outputs, not the machinery.

Coverage
We've seen 2 of ~4 submissions
Full eval: 0 Lite-only: 0 Unevaluated: 2
2 stories
1. The Pentagon Wanted a Master Key. Anthropic Said No. That Is Not the Story (github.com)
1 points by sheldonksalmon 2 days ago | 0 comments | skipped
2. Money Is the First AI – and We Never Noticed (github.com)
5 points by sheldonksalmon 3 days ago | 2 comments | skipped