This session examines how AI can be responsibly layered onto DPI to close current gaps such as enabling local-language access, conversational interfaces and inclusive decision-support across health, education and social protection. Multilingual large language models, voice interfaces and grounded public-data systems are treated not as add-ons but as the substrate that turns transactional infrastructure into genuinely public intelligence; systems a rural health worker or an ordinary citizen can actually use, in their own language, to get something done. The session presents a model that stimulates active AI development for public service delivery. The Nigeria AI Scaling Hub (NAISH) presents its model for moving AI solutions from pilot to public-service deployment, anchored on the Scaling AI for Development (SAID) Challenge; an instrument that matches problems articulated by Ministries, Departments and Agencies (MDAs) to operationally piloted Nigerian AI solutions.