Xiaomi has reorganised its AI around MiMo, its in-house base model, folding the XiaoAI assistant’s tech into three pillars: the MiMo model team led by Luo Fuli, cloud engineering under Luan Jian, and on-device OS work pushed into the phone business.
The move matters because MiMo is no longer an optional model. As Xiaomi’s ‘human-car-home’ ecosystem spans phones, cars, homes, wearables and robots, a single shared model base is what stops each product shipping its own disconnected assistant.
The strategic logic is about the entrance, not the benchmark. As AI agents take over tasks, users stop opening app stores and browsers; the agent calls services directly. A hardware maker without its own agent risks becoming a dumb chassis under someone else’s intelligent layer — exactly the data-ownership tug-of-war Xiaomi once had with Huami over wearable health data.
The cost is real: training, inference, talent and compute are a long-term drain, and phone makers are fundamentally hardware businesses. But if agent-to-agent collaboration becomes the norm, whoever controls the user entrance, the data and the service fulfilment holds the higher ground.
Xiaomi’s bet is that even if MiMo is never the largest or highest-scoring model, owning the agent that represents its ecosystem is worth more than the savings from renting someone else’s.
Original report: Leiphone.
Translated and adapted from Leiphone (leiphone.com).