Here’s what happened.
Generalist describes GEN-1.5 performing tasks after one demonstration, without additional training. The company reports an average success rate of 59% across the tasks in its evaluation. The same account describes fragility in the results, which matters when moving from a demonstration to repeatable work.
The We Are So Done take
First we wrote down how to do the job. Then we filmed ourselves doing it. Now the pitch is that one demonstration can be enough. Training your replacement is getting wonderfully efficient. The replacement still drops things.
For those of us already clicking
Watch the repeat attempts
- Access
- Vendor research; general product access has not been established.
- Start here
- Look at the task set, failures and repeated attempts. The reported 59% belongs to this evaluation, not to a profession.
- The small print
- The 59% figure is the vendor’s reported success rate in that test, not the share of workers replaced or our score. Product access for a typical visitor has not been established, and we have not independently reproduced the results.
Follow the evidence.
Source-based reporting. No independent hands-on test by this newsroom.
Sources checked 2026-09-23 · Editorial updated 2026-09-23