The AI agent metrics framework ft. Anastasia Zamyshlyaeva, Jeff Schinella & Jeremy Castile

Measuring the business impact and ROI of AI-powered engineering requires a structured framework centered on behavior, environment, and outcomes. While 96.4% of software teams have adopted AI coding tools, only 21% track specific metrics, creating a significant gap between perceived productivity and verifiable data. Effective measurement shifts from tracking simple lines of code or PR counts to analyzing agent autonomy, repository readiness, and actual production output. By transitioning from assistive to agent-native development, teams can achieve substantial velocity gains, provided they implement guardrails to ensure code quality and security. GitKraken Insights facilitates this transition by providing visibility into developer behavior, tool efficiency, and capital versus operational expenses. This data-driven approach allows engineering leaders to move beyond anecdotal evidence, optimize tool usage, and provide defensible proof of AI's value to stakeholders.