- Functional requirements that focus on what the agent will do and where intermediate humans will provide oversight.
- A set of non-functional requirements that focus on areas of performance, compliance, security, observability, and other operational requirements, just as they would for APIs and automations.
- Another set of non-functional requirements that focus on data, including data quality, governance, bias, and AI model maintenance.
Non-functional requirements for AI agents can be similar to those for applications, where user stories are granular and aim to deliver small, atomic functions. These NFRs can guide developers to respond as to develop functionality described in user stories and to help quantify what a code review should pass.
However, you may need another set of NFRs expressed at the feature or version level. These NFRs help qualify the release readiness of an AI agent, specify Data governance and AI requirements.and define others non-negotiable devops.
“For teams working with agent AI, it’s essential to differentiate which non-functional requirements are best enforced by machines, such as security, compliance, and scalability, and which still require human judgment, such as UX, aesthetics, and fast-feeling performance,” says Jonathan Zaleski, director of technical architecture at happyfuncorp. “The future of AI product development lies in hybrid workflows, where AI handles objective, measurable criteria at scale, and humans focus on the emergent and intuitive aspects that shape truly meaningful experiences.”
#write #nonfunctional #requirements #agents
