Format
Long (20 minutes)
Description
I built CS Brasil, a browser FPS that grew into a real product experiment: 2,191 players, 27 countries, live telemetry, bots, regression checks, and a game loop evolved with AI agents.
This talk is not "AI wrote my game". It is about the engineering system behind it: isolated builders, critics with clean context, bot simulation, invariants, browser checks, and the harness that decides whether a change is believable.
I will show how a browser game becomes a lab for agentic software development, where builders never grade their own work and every improvement must survive real evidence.
Takeaways:
- how to use agent loops without shipping plausible garbage
- why browser verification and invariants matter more than model confidence
- what game development teaches us about AI-assisted product engineering
Lien vers vos slides
No response
Autre
No response
Twitter
@rubenmarcusdev
Email
rubenmarcus.dev@gmail.com
Format
Long (20 minutes)
Description
I built CS Brasil, a browser FPS that grew into a real product experiment: 2,191 players, 27 countries, live telemetry, bots, regression checks, and a game loop evolved with AI agents.
This talk is not "AI wrote my game". It is about the engineering system behind it: isolated builders, critics with clean context, bot simulation, invariants, browser checks, and the harness that decides whether a change is believable.
I will show how a browser game becomes a lab for agentic software development, where builders never grade their own work and every improvement must survive real evidence.
Takeaways:
Lien vers vos slides
No response
Autre
No response
Twitter
@rubenmarcusdev
Email
rubenmarcus.dev@gmail.com