Cache frequent answers, smaller models for classification, truncate context, set max tokens, monitor spend alerts. ShopNest caches “shipping time” FAQ responses for 24h.
PwC rounds often probe: Enterprise AI.
Themes commonly reported in public MNC interview experiences (AmbitionBox / LinkedIn / Glassdoor-style). Practice aloud; keep answers this short in the real round.