AI Agent Usage in Software Engineering Doubles to 59%
- The adoption of agentic AI—artificial intelligence designed to act as agents capable of planning and executing tasks—is increasing within software engineering workflows.
- This figure represents nearly double the usage rate recorded in the organization's previous annual Developer Survey.
- Despite the rapid increase in adoption, the survey indicates that the implementation of agentic AI remains limited in scope.
The adoption of agentic AI—artificial intelligence designed to act as agents capable of planning and executing tasks—is increasing within software engineering workflows. According to a pulse survey published by the Stack Overflow Blog on May 27, 2026, usage of AI packaged as agents has reached 59%.
This figure represents nearly double the usage rate recorded in the organization’s previous annual Developer Survey.
Despite the rapid increase in adoption, the survey indicates that the implementation of agentic AI remains limited in scope. Most current professional applications rely on single-agent systems rather than multi-agent orchestrations.
The data further shows that these AI agents are generally monitored by developers while at work, suggesting a preference for human-in-the-loop oversight over full autonomy in software engineering environments.