Dev.to
6/29/2026

Two-day hackathon kicks off AI Engineer World’s Fair
Short summary
Two teams won a $35K hackathon at AI Engineer World's Fair showcasing advanced AI agents. SplatForge created a 93%-accurate object identifier that learns from Gemini's suggestions; Rote built a browser-automation agent that caches executable skills in MongoDB for instant global reuse, reducing token costs. Both exemplify the industry shift toward systems that improve themselves with minimal user input.
- •SplatForge: 93%-accurate object manipulation system that improves from Gemini suggestions
- •Rote: Browser-automation agent with MongoDB skill caching for global cost reduction and latency gains
- •Trend: Hackathons showcasing AI agents that learn and improve with minimal user input
Generated with AI, which can make mistakes.
Is this a good recommendation for you?



