Reproduce “A Zero-Shot Multi-Agent Framework for Human-Building Interaction via Programmatic Reasoning” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yuqi Wang, Gulai Shen, Ali Mehmani
Nantum AI and Harvard University researchers developed a zero-shot multi-agent framework that uses programmatic reasoning to enable intuitive Human-Building Interaction (HBI). The system interprets natural language queries to access real-time building data, documentation, and code, achieving 100% numerical accuracy and high data retrieval accuracy for complex building-related questions without domain-specific fine-tuning.
Key insights
- A zero-shot multi-agent architecture can effectively bridge natural language interaction with complex, specialized domain knowledge in HBI without extensive domain-specific fine-tuning.
- Programmatic reasoning, executed by specialized coding agents and interpreters, ensures precise arithmetic and logical calculations, mitigating LLM 'hallucinations' often seen in technical domains.
- The modular design allows for scalable integration of diverse LLMs, data sources, and tools, making the system adaptable to evolving building environments and user needs.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.