Generative design is one of the most oversold terms in current BIM marketing, and simultaneously one of the genuinely useful ones, provided the comparison stays honest about exactly what each approach is actually good for rather than treating generative design as a wholesale replacement for architectural judgment.
| Approach | Typical design options per week (early massing stage) | Best for |
|---|---|---|
| Traditional manual iteration | 3 to 6 options | Design intent requiring architectural judgment at every step |
| Generative design (parametric, rule-based) | Dozens to hundreds of variants, filtered by defined constraints | Optimising a well-defined problem (unit mix, solar orientation, structural grid efficiency) |
Where Generative Design Earns Its Place, and Where It Doesn't
It's genuinely strongest on well-defined optimisation problems - maximising saleable area within FSI limits, or testing structural grid options against a fixed cost model - where the evaluation criteria can be expressed clearly enough for software to iterate against automatically. It's weakest on genuinely creative or contextual design decisions, since a generative tool can readily produce a hundred massing options but can't judge which one actually respects a site's street character or a client's less quantifiable aesthetic priorities.
A Scenario Showing the Realistic Workflow
Picture a residential development team using generative design specifically to explore unit mix and orientation options against FSI and daylighting constraints, generating a hundred technically valid variants overnight. Rather than accepting any variant automatically, the architect reviews the shortlist the software narrows this down to, applying judgment about street character, entrance sequence and client preference that the software genuinely can't evaluate on its own - this narrowing-then-judging workflow is the realistic, valuable use of generative design on most Indian projects, not a replacement for the architect's final decision.