The survey revealed that 51% of respondents currently use AI agents in production, with mid-sized companies leading the way at 63%. Additionally, 78% of participants reported plans to implement agents in the near future.
Performance quality was cited as the most significant barrier to deploying agents, followed by concerns about safety and cost. Smaller companies emphasized quality more heavily, while larger enterprises prioritized safety due to regulatory requirements.
To address these challenges, companies are implementing controls such as tracing, offline evaluations, and guardrails. Larger enterprises tend to restrict agent permissions to read-only actions.
The survey also identified emerging themes, such as the need for better multi-agent collaboration and improved reasoning capabilities, highlighting key areas for future development.
Collection
[
|
...
]