“Most traditional hardware approaches are doomed,” says founder of Qiduo (奇朵)
The argument
Huang Yong (黄勇), founder of Qiduo (奇朵), argues that the vast majority of hardware teams are approaching AI-native products backward: they start by building a device and then tack AI on top. That, he says, is doomed to produce fragmented products. Instead, Huang insists you must design an intelligent agent first and then build a body for it — hardware becomes the agent’s anatomy, not merely a container. It has been reported that Huang once licensed APIs into nearly 20 million children’s devices, and industry peers in the Pearl River Delta reportedly call him “bangzhu” — a go-to troubleshooter across children’s hardware categories.
Product strategy
Qiduo’s debut — a science camera for 6–12 year olds powered by a multimodal “Little K teacher” agent trained on an expanded “100,000 Whys” knowledge base — illustrates the approach. Huang’s team removed screens and gimmicks, focused sensors and physical controls, and designed a generational product matrix: baby monitors for under-3s, cognitive cards for 3–6, the science camera for 6–12, and a Pocket-like DV for teens (think DJI (大疆) Pocket 3). Why chase the phone? Huang predicts a “unpacking” of smartphones as vertical devices re‑hardwire app features into specialized hardware that outperforms a generalist phone in specific scenarios.
Supply chain and geopolitics
Huaqiangbei’s rapid prototyping and China’s dense supply chain make fast iteration possible — teams can move from concept to shipment in months and Chinese booths reportedly dominated recent CES halls. But Huang also warns of new global compliance walls: export controls, sanctions and trade policy now shape which components and models can scale internationally. He says initial iterations will reuse mature supply lines, but true differentiation will require building new modules (for example, low‑light miniature camera modules) — both a challenge and a competitive moat.
What’s next?
Huang frames this as a structural industry shift: content formats become agents and hardware becomes the agent’s body. Qiduo is already thinking beyond cameras — toy robots and child‑sized machine dogs are on the whiteboard — and Huang believes a focused, stage‑by‑stage product matrix is more realistic than a single device “from 3 to 12.” Will the market reward agent‑first hardware? If supply, regulation and user behavior align, Huang argues it’s not only possible — it’s inevitable.
