Former DJI Engineer Builds Consumer Knitting Machine, Raises Millions

Clawlab (CLAWLAB) was founded in December 2022 by former DJI and Meituan engineer Hu Wenxin, who chose to tackle the largely untapped home textile‑machine category. Over the past three years the team deliberately stayed “under the radar,” inviting investors to the office for demo visits rather than conducting public roadshows.

Hu argues that textiles address the core human need of “clothing, food, shelter, and transportation,” and that after satisfying basic needs, consumers seek DIY customization as a next‑level expression. The DIY textile market globally encompasses over a hundred million potential users, yet it lacks a complete soft‑hardware solution; unlike 3D printing, there are no reusable parts or open‑source algorithms, and existing references are limited to legacy mechanical knitting machines.

In 2024 Clawlab launched an automatic tuft‑gun as a small‑scale overseas validation tool, generating nearly ¥100 million in revenue over two years. Building on that, the company introduced a consumer‑grade “Station” platform that eliminates the need for professional pattern making. Users can simply draw or photograph a design, and the system generates a knitting pattern that produces scarves, baby clothes, pet accessories, or small plush items.

The core technical challenge is not material handling but creating a real‑time perception, adjustment, and compensation control system for the highly sequential knitting process. Hu notes that different needles, yarn tensions, and garment shapes each require distinct mechanical actions and tension curves, which cannot be solved by post‑hoc tuning and must be defined in the system design from the start.

To overcome the lack of open algorithms and datasets, Clawlab spent nearly three years deconstructing the knitting process into programmable control algorithms, leveraging expertise from robot control, motion planning, and computer graphics. After a year and a half they achieved a reliable first‑generation knitting workflow.

Beyond hardware, the team developed a “textile AI Agent” – a vertical AI model enriched with domain knowledge that embeds their proprietary pattern‑generation algorithms. Users can upload an image or describe their design in natural language; the agent interprets the desired style, extracts features, and outputs a machine‑ready knitting pattern, compressing what traditionally takes days for a human pattern maker into minutes.

Clawlab has raised multiple rounds of financing exceeding several hundred million yuan, with investors including Sequoia Capital China, Shunwei Capital, Yuanjing Capital and miHoYo. The company plans to release a lightweight desktop‑level knitting product and build a content‑sharing community to serve hobbyists and micro‑customization needs.

Market analysis shows that active knitting enthusiasts in the West and East Asia number in the tens of millions, with broader potential exceeding one hundred million. Social platforms report billions of views on knitting‑related topics, indicating strong latent demand for easy‑to‑use, customizable textile creation.

Clawlab segments its target users into four groups: (1) heavy users of legacy knitting machines seeking efficiency and reliability; (2) small B‑to‑B operators (e.g., baby‑clothing makers, pet‑accessory creators) who need rapid, customizable production; (3) mainstream consumers without DIY experience who want instant, personalized textiles; and (4) future commercial users who will evolve from emotional consumption to lightweight commerce. Hu emphasizes that as users progress through these stages, the product must lower design barriers, maintain high quality, and eventually deliver fully customized finished goods.

Hu also highlights the company’s cultural philosophy: open communication, performance‑based compensation, and a focus on meaningful work rather than meaningless overtime. He believes that building a full‑stack ecosystem—hardware, software, AI, and community—creates a moat far wider than a single hardware product could achieve.