Example

A finished result

Nufar Gaspar's own run, October 1, 2026

This is the page the Benchmark Runner shows after the reveal, saved as it was at the end of the live Frontier Lab. Seven AI options answered five tasks from Nufar's working week. She scored all 35 answers with the names hidden, and an AI judge scored the same answers as a second opinion. Press Reveal the names to see who was who.

It is one person's tasks, one person's judgment, and one run, so it says what fits her work and nothing about yours. The notes are hers, dictated while scoring and lightly edited for reading. The companies, people, and figures in the tasks are made up, and the money amounts in the pricing task are shown in millions so that nobody reads them as real prices. Costs and times are what that run recorded on that day.

Keep going with us

Where to go from here

Cohort 6 · starts October 5

Executive Agent Leadership

Go deep on building and leading agents, and design your organization's agentic operating system.

Sign up

Cohort 4 · starts October 12

Executive Catch-Up

Need to catch up first? Get fluent in using AI and building with AI, for yourself.

Sign up
More from The AI Daily Brief