前保安杀进了AI决赛，高中生拿走25万！这AI比赛办得有点绝
Ranking
No observed public metrics; popularity remains neutral/archived.
Merged summary
TL;DR - TRAE’s AI creativity competition showcased how agentic coding tools are lowering software-development barriers, enabling participants from a former security guard to teenage students to build functional products. The strongest entries paired AI-assisted implementation with real-user testing, domain knowledge, and iterative product decisions.
- The winning “词元开物” system combines an AI hardware IDE with a desktop PCB engraver, automating requirements, component selection, code generation, simulation, PCB layout, and fabrication.
- Runner-up MotionFrame uses iPhone, Apple Watch, and AirPods data to analyze badminton technique and deliver coaching; its high-school creators launched it on Apple’s App Store.
- TRAE’s SOLO, Skills, Rules, MCP, custom agents, Plan, and Spec features let AI decompose goals, use tools, edit files, run commands, and debug projects.
- Across 37,000 applicants and 14,000 entries, successful teams moved beyond demos by validating with users, grounding AI outputs in real samples, and repeatedly fixing deployment and hardware-integration failures.
Sources (1)
前保安杀进了AI决赛，高中生拿走25万！这AI比赛办得有点绝
TL;DR - TRAE’s AI creativity competition showcased how agentic coding tools are lowering software-development barriers, enabling participants from a former security guard to teenage students to build functional products. The strongest entries paired AI-assisted implementation with real-user testing, domain knowledge, and iterative product decisions.
- The winning “词元开物” system combines an AI hardware IDE with a desktop PCB engraver, automating requirements, component selection, code generation, simulation, PCB layout, and fabrication.
- Runner-up MotionFrame uses iPhone, Apple Watch, and AirPods data to analyze badminton technique and deliver coaching; its high-school creators launched it on Apple’s App Store.
- TRAE’s SOLO, Skills, Rules, MCP, custom agents, Plan, and Spec features let AI decompose goals, use tools, edit files, run commands, and debug projects.
- Across 37,000 applicants and 14,000 entries, successful teams moved beyond demos by validating with users, grounding AI outputs in real samples, and repeatedly fixing deployment and hardware-integration failures.