Research
What we're learning about agents doing real work.
- Verification
- How do we know an agent actually accomplished an objective?
- Reliability
- What predicts failure in long-running agentic systems?
- Deployment
- What changes when agents encounter real businesses?