Localintent enhances smart building automation outcomes with local llms
LocalIntent: Outcome-Specific Structured Generation for Backend-Compatible Smart Building Automation with Local LLMs
Artificial Intelligence
Summary
Smart building systems need to understand and act on voice or text commands correctly. The authors studied a way to improve how small local language models handle these commands by tailoring prompts depending on the type of task. Their method, LocalIntent, helps the system better understand instructions, pick the right devices, and generate valid actions, especially for complex rule-based commands. This means smarter buildings can run more reliably without sending data to cloud servers.
What this means in practice
- •For smart building developers: Implement local language models that better recognize user intents and generate valid control commands for building devices, improving automation reliability without cloud dependency.
- •For home automation integrators: Use outcome-specific prompting strategies to produce accurate rule-based automation scripts that align with local system capabilities for in-home smart device control.
Authors
Tianyang Liu, Jiayue Wang, Zhanhao Cheng, Changyang Wang, Shaoshu Zhu, Zhenghao Wu, Hadi Tabatabaee Malazi
Abstract
Locally deployed large language models (LLMs) can reduce reliance on remote cloud services for natural language smart building automation, but backend-facing processing requires correct outcome selection, alignment with local capabilities, and valid structured outputs. We investigate whether a complete outcome-specific prompting design improves compact-model outputs beyond generic route-then-generate prompting. The proposed LocalIntent approach routes requests to ordinary or persistent-rule branches: the former produces commands, clarifications, or rejections, while the latter generates trigger-action rules, followed by deterministic validation. We evaluate four locally executed Gemma and Qwen models spanning 2B-9B parameters on 101 non-persistent and 54 persistent-rule records, comparing Monolithic prompting, generic two-stage prompting, LocalIntent, and validation-guided repair. Across the evaluated datasets, LocalIntent achieves higher intent understanding, device identification, and action/parameter correctness than generic two-stage prompting, while clarification/mode correctness remains model-dependent. The largest gains occur for rules, where LocalIntent improves intent understanding from 26.4% to 80.6%, device identification from 59.3% to 84.7%, rule-mode correctness from 49.1% to 98.6%, and action/parameter correctness from 5.6% to 44.9%. For non-persistent requests, pooled gains are smaller: intent understanding increases from 39.6% to 53.2%, and action/parameter correctness from 30.7% to 42.8%.