OPEN LINK ↗
// 13m agoPRODUCT LAUNCH
Tokyo Now debuts real-time AI pixel map
Built by Netlify engineer Elad Rosenheim, the application autonomously aggregates local listings, social posts, and transit feeds every five minutes, using LLMs to extract live events and push changes instantly over WebSockets. The entire architecture runs on Netlify Server, Netlify Database, and Netlify's AI Gateway without external backend infrastructure.
// ANALYSIS
Tokyo Now showcases the emerging pattern of autonomous full-stack agents, demonstrating how long-running server backends and managed AI gateways can replace fragmented infrastructure.
- –Full-stack agent runtime: Netlify Server maintains a persistent Node.js process with scheduled wakeups and WebSocket connections, streaming changes to browser clients within seconds.
- –Multi-model pipeline: Combines Gemini 3.1 Flash Lite for event extraction and deduplication, Gemini Flash Image for pixel art sprites, and Grok and Qwen via OpenRouter for live search.
- –Budget-aware execution: Employs a paced daily AI budget with cost reservations, throttling background tasks during quiet hours and scaling capacity when multiple events trend.
- –Contextual sensor fusion: Verifies online event claims against real-world signals, including Docomo bike-share station anomalies, weather changes, and Haneda flight data.
// TAGS
tokyo-nownetlifyagentstreamingimage-gensearchcloud
DISCOVERED
13m ago
2026-11-13
PUBLISHED
40m ago
2026-11-13
RELEVANCE
7/ 10
AUTHOR
Netlify