This paper proposes a multi-agent development paradigm in which formal methods—specifically VDM (Vienna Development Method)—serve as rigorous contracts between AI agents. Test-Driven Development is fundamentally limited by its grounding in inductive reasoning; formal specifications provide deductive guarantees suitable for inter-agent coordination. We present a workflow in which a human domain expert defines requirements through structured dialogue, and AI agents autonomously derive specifications, architecture, and implementation, with the formal specification functioning as the team's common language. The human role is redefined—from "the person who writes code" to "the person who defines what to build and why, and evaluates whether AI's proposals match business intent." Natural-language specifications produce exponentially growing interpretive discrepancies as the number of coordinating agents increases; formal specifications keep module boundary consistency mechanically verifiable regardless of agent count, making the approach inherently scalable.
Hikaru Ando (Thu,) studied this question.