Xiaomi Miloco 2.0: Open-Source AI That Makes Homes Remember, Recognize People, and Act
Xiaomi has officially released and open-sourced its forward-looking "All-Household Smart AI Open Source Solution" — Xiaomi Miloco 2.0. Built on Xiaomi's proprietary MiMo large model, the solution is integrated into the OpenClaw framework as an Agent, delivering a comprehensive upgrade across interaction methods, product features, and memory systems.
Unlike the first-generation Miloco, version 2.0 has undergone a fundamental reconstruction of its underlying logic, evolving from traditional rule-based triggers into a family smart agent with autonomous reasoning capabilities. This marks a true entry into a new stage of proactive intelligence, defined by "memory, recognition, and execution." The code is fully open-sourced on GitHub.

Four Core Capabilities That Redefine Home Intelligence
Miloco 2.0 introduces four core capabilities: general common sense, identity recognition, family memory, and family tasks. Without any pre-set rules, the system can automatically detect potential hazards and issue tiered alerts — for example, reminding when children are playing with knives or when an elderly person falls. Additionally, it integrates facial and posture data, enabling the large model to identify family members, automatically register new members, and deliver personalized interactions.
The family memory function gathers long-term habits and preferences from daily interactions, providing a reference for the Agent's decision-making. Stable habits can be upgraded into automated family tasks, moving beyond simple conditional triggers like "turn on the lights when someone enters" to more complex scenarios such as scheduled medication reminders and daily exercise tracking — tasks the Agent can complete autonomously after understanding the intent.
Proactive Intelligence Brings a Butler into Your Home
Building on these four capabilities, Miloco 2.0 delivers "proactive intelligence" — the system behaves like a butler who is knowledgeable, family-aware, and good at planning, completing tasks before users even issue a command. In a demonstration scenario, when the male owner comes home later than usual, the system infers he may have worked overtime, and the Xiaoai speaker automatically plays a caring voice message: "It's tiring working late, make sure to rest early."
For deployment, the solution requires at least 4GB of memory and 256GB of storage. A Mac mini running 24/7 is recommended, and both macOS and Linux are supported. Windows users need to run it through WSL. You'll also need a Xiaomi account, devices already connected to Mi Home, and a multimodal large model API key. The official recommendation is to use the MiMo-v2.5 series model. Additionally, the accompanying Web home panel lets you view the home overview, device status, family member profiles, and historical event logs in real time, giving you full control over the entire smart home system.
Related article
DeepMind CEO Hassabis: I sleep six hours a day, usually feel energetic around 1 a.m.
Fortune recently featured an interview with Demis Hassabis, CEO of Google DeepMind, revealing his unconventional approach to rest and productivity. Hassabis disclosed that he sleeps very little, structuring his waking hours into two distinct work blo
OpenAI, Anthropic Vie for Market Share Despite Revenue Shortfalls
Despite recent reports suggesting OpenAI missed revenue targets, creating pressure on tech stocks this Tuesday, private AI lab investors remain resilient. Seasoned backers have confirmed they will not reduce investment despite negative media coverage
California AV Compliance: A New Era of Tickets, Geofences, and 1M Miles
Guident operates an AuveTech shuttle in South Florida, managing a four-mile route in West Palm Beach and a one-mile route in Boca Raton using its remote monitoring technology. | Credit: GuidentCalifornia is redefining the regulatory landscape for dri
Related Special Topic Recommendations
Comments (0)
0/500
Xiaomi has officially released and open-sourced its forward-looking "All-Household Smart AI Open Source Solution" — Xiaomi Miloco 2.0. Built on Xiaomi's proprietary MiMo large model, the solution is integrated into the OpenClaw framework as an Agent, delivering a comprehensive upgrade across interaction methods, product features, and memory systems.
Unlike the first-generation Miloco, version 2.0 has undergone a fundamental reconstruction of its underlying logic, evolving from traditional rule-based triggers into a family smart agent with autonomous reasoning capabilities. This marks a true entry into a new stage of proactive intelligence, defined by "memory, recognition, and execution." The code is fully open-sourced on GitHub.

Four Core Capabilities That Redefine Home Intelligence
Miloco 2.0 introduces four core capabilities: general common sense, identity recognition, family memory, and family tasks. Without any pre-set rules, the system can automatically detect potential hazards and issue tiered alerts — for example, reminding when children are playing with knives or when an elderly person falls. Additionally, it integrates facial and posture data, enabling the large model to identify family members, automatically register new members, and deliver personalized interactions.
The family memory function gathers long-term habits and preferences from daily interactions, providing a reference for the Agent's decision-making. Stable habits can be upgraded into automated family tasks, moving beyond simple conditional triggers like "turn on the lights when someone enters" to more complex scenarios such as scheduled medication reminders and daily exercise tracking — tasks the Agent can complete autonomously after understanding the intent.
Proactive Intelligence Brings a Butler into Your Home
Building on these four capabilities, Miloco 2.0 delivers "proactive intelligence" — the system behaves like a butler who is knowledgeable, family-aware, and good at planning, completing tasks before users even issue a command. In a demonstration scenario, when the male owner comes home later than usual, the system infers he may have worked overtime, and the Xiaoai speaker automatically plays a caring voice message: "It's tiring working late, make sure to rest early."
For deployment, the solution requires at least 4GB of memory and 256GB of storage. A Mac mini running 24/7 is recommended, and both macOS and Linux are supported. Windows users need to run it through WSL. You'll also need a Xiaomi account, devices already connected to Mi Home, and a multimodal large model API key. The official recommendation is to use the MiMo-v2.5 series model. Additionally, the accompanying Web home panel lets you view the home overview, device status, family member profiles, and historical event logs in real time, giving you full control over the entire smart home system.
DeepMind CEO Hassabis: I sleep six hours a day, usually feel energetic around 1 a.m.
Fortune recently featured an interview with Demis Hassabis, CEO of Google DeepMind, revealing his unconventional approach to rest and productivity. Hassabis disclosed that he sleeps very little, structuring his waking hours into two distinct work blo
OpenAI, Anthropic Vie for Market Share Despite Revenue Shortfalls
Despite recent reports suggesting OpenAI missed revenue targets, creating pressure on tech stocks this Tuesday, private AI lab investors remain resilient. Seasoned backers have confirmed they will not reduce investment despite negative media coverage





Home






