Enterprise Tool Use
How well agents complete real enterprise workflows across 11 industries, using simulated CRM, ERP and ticketing tools, graded on whether the task actually got done and how it got there.
11
Industries
26
Models evaluated
[date]
Prices recorded
Agentic intelligence index
Composite pass^5 score across Tool use evaluations (higher is better).
Error bars show 95% confidence intervals. We test models on private, non-contaminated tasks.
RL Gyms and evaluation data
License tasks and environments for training and evaluating your models. Request samples to see the format and content.
Methodology
Private tasks, rebuilt in full
Test tasks are never published. We rebuild each set rather than patch it. The current Enterprise Tool Use set went live on September 29.
RL Gyms and evaluation data
License tasks and environments for training and evaluating your models. Request samples to see the format and content.