Live site · found by @thomaskanze
AI Subway Disruption Simulator for New York and London
Test an AI controller by disrupting simulated New York subway or London Underground networks
Jev judged this in 274 ms for $0.00009
Found something worth sharing?
Drop the link and Jev reads it on the spot. What belongs lands in the community feed with your name on it.
More from the community
no
@0x4e53
Jev is insane ✨ On Saturday, I built a harness to play Overcooked. A planning model builds a live state / understanding of the level, and Jev agents make decisions in real time! Coding agents can review experiment data and improve the harness. #jev #opus #ai #gameai
LinkedInGames and real time
Chef Jev AI agent plays Overcooked in real time
An AI agent that plays Overcooked in real time, using a fast decision model and a high-level planner to coordinate four players
Jev · 408ms@0x4e53
Nishit Khalpada
@nishitkhalpada
I wanted to understand where Jev actually helps, so I built a small traffic simulation with GPT-6 Astra handling the build and Jev making the traffic decisions. One side uses a normal fixed 8-second signal cycle. The other side lets Jev decide whether to keep the current green or switch based on queue sizes, wait times, recent arrivals, and traffic trends. Both sides get the exact same cars and follow the same movement rules. What I found interesting is that Jev doesn’t always win. Fixed timing works really well when traffic is balanced. Jev starts doing better when traffic becomes uneven or keeps changing. The video is the heavy-traffic run. I’ll add the Light, Moderate, and Heavy results, along with the code, in the replies.
XGames and real time
Adaptive Traffic Signal Control vs Fixed Timing Simulation
Jev · 261ms@nishitkhalpada
