• In-house model development: HDFC Bank developed its proprietary AI platform, Neve, with a 40-person deep-tech team on an initial budget of around 2 crore INR, prioritizing accurate, domain-specific smaller models over costly large language model token consumption.
• Small language models: Wipro focuses on micro-small, specialized domain models to reduce latency and improve response speed, which the company views as critical for large-scale enterprise adoption.
• Agentic security risks: Palo Alto Networks highlighted that data exfiltration times have dropped to under 25 minutes, requiring continuous zero-trust verification for ephemeral agents alongside consolidated enterprise data to preempt attacks.
• Workflow automation adoption: Zepto is deploying agents across procurement, dark-store supply chain planning, and vendor invoice processing to transition business-as-usual operations into an automated flow where humans only manage exceptions.
• Workforce impact framework: Wipro categorizes roles into four quadrants—replace, augment, eliminate, and AI-only—while aggressively hiring AI-native campus recruits and reskilling middle management.