Heimstaden Scales OpenAI Codex as NVIDIA Expands Smart City AI

BARCELONA & SAN FRANCISCO — Enterprise property management, municipal infrastructure, and consumer wearables marked a series of artificial intelligence deployments on Tuesday, led by residential real estate operator Heimstaden scaling OpenAI tools across its European portfolio and NVIDIA expanding Physical and Agentic AI services ahead of the Smart City Expo World Congress in Barcelona (November 3 to 5).

Heimstaden Scales ChatGPT Enterprise, Codex, and Amplio Estate Across European Real Estate

According to OpenAI News, European residential real estate operator Heimstaden reached 50% employee adoption of ChatGPT Enterprise within 2 to 3 months of launching an AI First Team in 2025 with full executive sponsorship from Co-CEO Helge Krogsbøl and Chief Digital Officer Nikolaj Brammer.

In the Netherlands, where approximately 25% of rental properties require virtual furnishing when listed, Heimstaden replaced an external agency charging 50 euros per staged image and taking 2 to 3 days of turnaround with an internal AI image workflow built in 2 weeks by 1 developer. That deployment cut per-image staging costs by more than 99%—to less than 0.10 euros per photo—saving 130,000 euros annually while allowing letting teams to publish listings within hours.

Meanwhile, Environmental Sustainability Analyst Julia Erlandsen used OpenAI Codex to process climate risk data across 20,112 assets and more than 6 million rows of unstructured records. A data integration project originally estimated to require 5 people working for 1 year was completed by 2 people in 2 months. Manager of the AI First Team Andreas Kalmar Lindquist also oversaw the launch of Amplio Estate, a commercialized platform built on the OpenAI API and Model Context Protocol (MCP) that structures tenant communications, procurement, and maintenance workflows across 80% of repetitive administrative tasks.

NVIDIA Expands Physical and Agentic AI for Global Smart Cities in Barcelona

Ahead of the Smart City Expo World Congress in Barcelona running November 3 to 5, NVIDIA Blog announced that municipal agencies and infrastructure operators across Europe, Asia, Africa, and North America are deploying NVIDIA Omniverse libraries, NVIDIA Cosmos world foundation models, NVIDIA NeMo, NVIDIA NIM microservices, and the NVIDIA Metropolis platform:

• Spain, United States & Poland Agentic Public Services: In Spain, 1MillionBot integrated sovereign model ALIA and NVIDIA Nemotron across more than 150 municipal deployments in Madrid and Catalonia to provide 24/7 multilingual constituent assistance. In the San Francisco Bay Area, LiveX deployed digital service agents on NVIDIA RTX PRO GPUs to reduce 311 call center wait times. In Wrocław, Poland, K2K built an NVIDIA NeMo and NIM platform unifying transport, waste, lighting, and acoustic monitoring.
• Malaysia, Nigeria, Taiwan & Morocco Digital Twins: In Malaysia, ITMAX is scaling its SamurAI V2 video analytics platform to more than 17,000 cameras by 2027, while its TeraSim traffic twin reduced urban travel times by 26%. In Nigeria, Honeywell is deploying its Forge Plant Cognition suite with NVIDIA Omniverse and Metropolis at the Dangote refinery in Lekki. In Taiwan, Linker Vision and Dutch architecture firm MVRDV combined Cosmos and Omniverse to simulate 15-year urban forestry outcomes in 10 minutes. In Morocco, Cires Technologies deployed Metropolis across Tanger Med, the largest port in the Mediterranean and Africa.

Apple Watch Audio Intelligence Enters iOS 27.3 Beta; Google Expands Dreambeans to India and AO3 Clone Shuts Down

Three consumer AI and copyright developments also surfaced on Tuesday morning:

• Apple Watch Series 12 and Ultra 4 Audio Intelligence Beta: As reported by Redmond Pie, Apple began developer beta testing in iOS 27.3 for Audio Intelligence, an opt-in ambient audio suite exclusive to Apple Watch Series 12 and Apple Watch Ultra 4 powered by the S11 chip's neural engine and Private Cloud Compute. The suite includes Live Rewind, which buffers 15-second ambient audio snippets in the S11 secure enclave and deletes them automatically after 30 seconds, and Siri Recaps, which generates encrypted bullet-point conversation summaries deleted after 7 days while displaying an orange indicator dot on the watch face whenever active.
• Google Expands Dreambeans AI Story App to India in 13 Languages: According to TechCrunch, Google expanded its experimental Gemini-powered personalized storybook and Veo short-video app Dreambeans to India on Android and iOS after recording approximately 55,000 U.S. downloads since its June launch. Product manager Gozde Oznur confirmed that the app now supports 13 Indian languages—including Hindi, Tamil, Telugu, Bengali, and Marathi—and lets users optionally connect Gmail, Google Photos, Calendar, Search, and YouTube context to generate 2 daily storybooks and 1 daily video.
• Vibe-Coded AO3 Clone Shuts Down After 2,000 Takedown Notices: Meanwhile, Dexerto reported that ao3.life, an unauthorized mirror launched across 40 virtual machines during a 4-week maintenance outage at fanfiction repository Archive of Our Own (AO3, which hosts 18.5 million works), shut down after drawing 126,592 unique users in 24 hours, selling 11 AI-narrated audiobooks, and receiving more than 2,000 author takedown notices in a single day.