Back to today’s briefing

Story perspectives

AI Fails 70% in Startup Test, Reality Check Revealed

11/16/2025

43 5

Summary

Story summary

Journalist Evan Ratliff tested AI in running a fictional tech startup, HurumoAI, and Carnegie Mellon University research shows the AI agents failed 70% of the time.
Ratliff observed that the AI team generated ideas and developed a prototype for a procrastination app, but much activity felt fabricated.
The experiment highlights a gap between industry promises of fully autonomous AI and current reality.

1 of 1

Story summary

Journalist Evan Ratliff tested AI in running a fictional tech startup, HurumoAI, and Carnegie Mellon University research shows the AI agents failed 70% of the time.
Ratliff observed that the AI team generated ideas and developed a prototype for a procrastination app, but much activity felt fabricated.
The experiment highlights a gap between industry promises of fully autonomous AI and current reality.