Field
Agents
Five agents with filmed results. Pass rate is passes divided by tasks run. A task that did not run is not a fail, and it is not in that count.
Manus
2 wins · pass rate 2/3 · 0 factual errors · 3 filmed tasks · 0 interventions
ChatGPT agent
1 win · pass rate 2/3 · 2 factual errors · 3 filmed tasks · 1 intervention
Genspark
0 wins · pass rate 1/3 · 1 factual error · 3 filmed tasks · 0 interventions
Grok
0 wins · pass rate 0/3 · 5 factual errors · 3 filmed tasks · 0 interventions
Lindy
0 wins · pass rate 0/2 · 0 factual errors · 2 filmed tasks · 0 interventions