# HomeFlow: A Data Flywheel for Smart Home Agent Training with Verifiable Simulation

arXiv: https://arxiv.org/abs/2606.01230 · HTML: https://mlsanity.com/paper/2606.01230/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Financial trading and forecasting agents, one of 16 directions in LLM agents and tool use.
- Builds on 21 of its 34 references that are in the map.
- First posted on arXiv 31 May 2026.

## Placed in

- LLM agents and tool use / [Financial trading and forecasting agents](https://mlsanity.com/agents/financial-trading-and-forecasting-agents/)

## Builds on (most cited first, 21 of 21)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [SWE-bench: Can Language Models Resolve Real-World GitHub Issues?](https://mlsanity.com/paper/2310.06770/) (2023)
- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [Self-Instruct: Aligning Language Models with Self-Generated Instructions](https://mlsanity.com/paper/2212.10560/) (2022)
- [The Rise and Potential of Large Language Model Based Agents: A Survey](https://mlsanity.com/paper/2309.07864/) (2023)
- [A Survey on LLM-as-a-Judge](https://mlsanity.com/paper/2411.15594/) (2024)
- [WizardLM: Empowering large pre-trained language models to follow complex instructions](https://mlsanity.com/paper/2304.12244/) (2023)
- [GLM-5: from Vibe Coding to Agentic Engineering](https://mlsanity.com/paper/2602.15763/) (2026)
- [LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods](https://mlsanity.com/paper/2412.05579/) (2024)
- [Character-LLM: A Trainable Agent for Role-Playing](https://mlsanity.com/paper/2310.10158/) (2023)
- [WebAgent-R1: Training Web Agents via End-to-End Multi-Turn Reinforcement Learning](https://mlsanity.com/paper/2505.16421/) (2025)
- [The Oscars of AI Theater: A Survey on Role-Playing with Language Models](https://mlsanity.com/paper/2407.11484/) (2024)
- [Synthetic Dialogue Dataset Generation using LLM Agents](https://mlsanity.com/paper/2401.17461/) (2024)
- [WebSynthesis: World-Model-Guided MCTS for Efficient WebUI-Trajectory Synthesis](https://mlsanity.com/paper/2507.04370/) (2025)
- [A Survey on LLM-based Conversational User Simulation](https://mlsanity.com/paper/2604.24977/) (2026)
- [HomeBench: Evaluating LLMs in Smart Homes with Valid and Invalid Instructions Across Single and Multiple Devices](https://mlsanity.com/paper/2505.19628/) (2025)
- [SimuHome: A Temporal- and Environment-Aware Benchmark for Smart Home LLM Agents](https://mlsanity.com/paper/2509.24282/) (2025)
- [SAGE: Smart home Agent with Grounded Execution](https://mlsanity.com/paper/2311.00772/) (2023)
- [On-Device LLMs for Home Assistant: Dual Role in Intent Detection and Response Generation](https://mlsanity.com/paper/2502.12923/) (2025)
- [Leveraging LLMs for Efficient and Personalized Smart Home Automation](https://mlsanity.com/paper/2601.04680/) (2026)
- [HearthNet: Edge Multi-Agent Orchestration for Smart Homes](https://mlsanity.com/paper/2604.09618/) (2026)

## Built on by (most cited first, 2 of 2)

- [DevicesWorld: Benchmarking Cross-Device Agents in Heterogeneous Environments](https://mlsanity.com/paper/2607.13465/) (2026)
- [SimVerity: When Does Simulated Agent Success Survive Physical Deployment?](https://mlsanity.com/paper/2608.25067/) (2026)

## Abstract (arXiv)

Large language model agents are moving beyond text-only interaction toward physical-world control, with smart homes as a representative domain. Real domestic interaction requires understanding ambiguous intents, operating in dynamic environments, and performing multi-turn reasoning. However, existing methods struggle to generate high-quality training data for smart home agents. We propose HomeFlow, a verifiable data flywheel for this domain. HomeFlow uses HomeEnv as a unified simulation environment and HomeMaker to procedurally generate diverse home settings. Subsequently, Blueprint compiles open-ended user intents into executable state-based success conditions, while MCTS-Flow synthesizes diverse, verifiable multi-turn trajectories through environment-guided tree search. We then optimize the agents via supervised fine-tuning and step-wise RLVE, which facilitates iterative improvement through authentic physical feedback. We further construct SmartHome-Bench to evaluate the agent across various smart home tasks. On this benchmark, HomeFlow-RL-4B and HomeFlow-RL-8B achieve task success rates of 84.60% and 87.03%. It is worth noting that HomeFlow-RL-8B even surpasses the leading GPT-5.5 by 1.23 percentage points.