Tag: expert fidelity

Esther Dyson, founding chair of ICANN, brought her electrolyte and GLP-1 questions to Ashley Koff's onix, then phoned the real Ashley to compare.

People asked ChatGPT, Claude, and Gemini what Ashley Koff, RD recommends. It answered in her voice, from her public work, with her clinical judgment left out.

Frontier models scraped every word experts ever published, and it still is not enough: the judgment that makes someone worth asking was never written down. Built from an expert's consented corpus, our onixes already represent a specific human more faithfully than the strongest frontier models. The expert in the loop is the moat no scrape can reach.

Every major AI lab publishes benchmarks for reasoning, coding, and math. But nobody is measuring whether AI actually gets human expertise right. We built an evaluation framework to find out.