Cities Worldwide Build and Scale AI Services With NVIDIA Physical and Agentic AI
NVIDIA is extending physical and agentic AI services to cities worldwide, leveraging Nemotron LLMs, Cosmos world models, and Metropolis video analytics on RTX PRO GPUs to enable resident‑service bots, vision agents, and digital twins. Deployments such as 1MillionBot in Spain and ITMAX’s SamurAI V2 in Malaysia illustrate how modest GPU workstations can power thousands of camera feeds and municipal procedures, with traffic‑signal simulations cutting peak‑hour travel times by up to 26%. This shared, GPU‑accelerated infrastructure lets municipal departments launch AI‑driven services without large‑scale compute investments, accelerating public‑service responsiveness and infrastructure planning.
⚡ Key Takeaways
- 1MillionBot handles municipal procedures, appointment requests, urban‑planning incidents and service inquiries across 150+ deployments in Spain using open‑source ALIA models and NVIDIA Nemotron.
- Vision AI agents in Wrocław and Malaysia use NVIDIA Cosmos world reasoning models and Metropolis Blueprint for video search and summarization (VSS) on RTX PRO GPUs.
- ITMAX’s SamurAI V2 vision agents will support 17,000+ cameras by 2027, running on NVIDIA RTX PRO 6000 GPUs and integrating the open‑source TeraSim traffic simulation platform.
- Traffic‑signal simulation with TeraSim reduces peak‑hour corridor travel times by up to 26%.
- Deployments can start on a single NVIDIA RTX workstation or a small GPU system rather than a massive data‑center.
Engineers can rapidly prototype city‑scale AI services using off‑the‑shelf NVIDIA GPUs and pre‑built models, achieving measurable operational gains (e.g., 26% travel‑time reduction) without the overhead of building dedicated compute clusters.
✅ Practical Steps
- Deploy resident‑service agents by integrating open‑source ALIA models with NVIDIA Nemotron on an NVIDIA RTX workstation as demonstrated by 1MillionBot.
- Build vision AI pipelines using NVIDIA Cosmos world models and Metropolis VSS Blueprint on RTX PRO GPUs, following the Wrocław and ITMAX implementations.
- Incorporate TeraSim traffic simulation into your vision‑agent stack to evaluate signal timing and achieve travel‑time reductions similar to ITMAX’s 26% improvement.
Want the full story? Read the original article.
Read on NVIDIA Blog ↗