In the 1990s, RAD was the rebellion against multi-year waterfall programs: short cycles, joint application design sessions, evolutionary prototypes, and component reuse. Critics said it produced brittle systems and weak architecture. Advocates said it produced working software while committees were still arguing over requirements documents.
Both were sometimes right. What changed in 2024–2026 is not the theory — it is the marginal cost of producing a credible prototype.
What classic RAD actually required
- Business users available for workshops — not a proxy product manager three layers away.
- Tooling that could assemble UI and data paths quickly (4GLs, early visual builders).
- Timeboxes that forced scope cuts instead of infinite polish.
- A reuse library so each prototype did not invent persistence and auth again.
Why the comeback thesis holds
Modern “vibe” sessions look like RAD workshops with a tireless scribe. Cloud agents look like overnight prototype factories. Component systems, design tokens, and managed backends look like the reuse libraries RAD manuals begged for.
But the failure modes return too. Unreviewed prototypes become production. Speed without architecture becomes rewrite debt. User workshops without decision rights become theater. The comeback only works if teams pair RAD energy with go/no-go gates borrowed from more ceremonial methods.
A practical RAD-for-agents checklist
- Name the timebox and the disposable artifact before you generate.
- Require a human workshop checkpoint before promoting a prototype path to a shared branch.
- Invest in the reuse layer (auth, observability, design system) so speed compounds.
- Write the abort criteria: what evidence kills the prototype?
- Graduate survivors into requirements-backed hardening — do not pretend the demo was the design.
Place on the memory lane
Waterfall still wins when the interface is physics. RUP still wins when risk must be staged. Agile still wins when priorities churn weekly. RAD wins when discovery is the bottleneck and prototypes are cheap. The industry’s mistake was declaring one method the permanent winner. The publication’s job is to notice when the cost curves flip.