Meet the winners of our Claude Opus 4.8 Build Day hackathon
Summary
The Claude Opus 4.8 Build Day hackathon, held on June 13 in San Francisco, gathered 310 founders and builders who, with \$500 in credits, developed working demos in 12 hours. First place went to Tekton, created by Holly Tang and Austin Burgess, a 3D reconstruction platform that revives Tang Dynasty architecture, tracing every component to historical sources using Opus 4.8 for verification and self-correction. Second place winner Sim Francisco, by Tanmayi Priya Dasari and Tejas Prabhune, is a digital twin of San Francisco's population with 10,000 synthetic residents, capable of polling the city and forecasting real-world outcomes with high accuracy; Opus 4.8 built and verified its backend. Custom Universe, by Jake Stevens and Mauricio Pereira, secured third place, offering a real-time engine that converts phone photos into editable, photorealistic 3D scenes for robotics training data, with Opus 4.8 assisting in development and tool selection.
Key takeaway
For AI Engineers and entrepreneurs developing complex applications, these hackathon projects demonstrate Claude Opus 4.8's capability to accelerate development and verification. You should consider leveraging large language models not just for code generation but also for comprehensive project planning, tool selection, and optimizing expensive operations like inference. This approach can significantly reduce development time and costs, enabling rapid prototyping of advanced systems for diverse fields from cultural preservation to robotics and demographic forecasting.
Key insights
Claude Opus 4.8 accelerates complex AI application development, enabling advanced 3D reconstruction, synthetic population modeling, and real-time 3D scene creation.
Principles
- Map the whole project before building.
- Optimize expensive inference calls early.
- Use LLMs for tool selection, not just code.
Method
Claude Opus 4.8 was used to research, assemble, verify, and optimize complex AI systems, including generating code, operating remote GPUs, and integrating diverse technologies.
In practice
- Reconstruct historical buildings with evidence chains.
- Poll synthetic populations for real-world forecasts.
- Generate synthetic data for robotics training.
Topics
- Claude Opus 4.8
- AI Application Development
- 3D Reconstruction
- Digital Twins
- Synthetic Data
- Robotics Simulation
Code references
Best for: AI Engineer, Machine Learning Engineer, Entrepreneur
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by Claude Blog.