- Published on
Hardware Modding Is the First AI Dividend for Ordinary People
- Authors
- Name
- DeepHorus
Recently, more and more ordinary users have begun using AI Agents to “radically mod” hardware they already own. From flashing Android onto Kindles to install WeChat Reading, to removing ads from Xiaomi TVs, device modification is moving from professional operations toward attempts by the general public. AI no longer only writes copy or makes images; it has begun directly getting hands-on with specific device problems.
A user used WorkBuddy to flash Android onto a Kindle and install WeChat Reading; the entire process involved only entering a URL on the device and pressing Enter. Jailbreaking, file copying, and other steps were all completed by the AI, with the user mainly acting as supervisor. Five or six years ago, the same operation required mailing the Kindle to a merchant and spending nearly 200 yuan.
On October 17, developer Gathin sent a Xiaomi TV model to Claude Code and had it connect to the TV over the same Wi-Fi. It disabled 25 Xiaomi system components and wrote a minimalist launcher, so that after boot the TV could go directly into commonly used apps. The before-and-after comparison posted on X received millions of views; CleanYourTV accumulated tens of thousands of stars in a little over ten days, and the community list covers more than 80 brands and over 300 TV models.
Similar cases are increasing. On October 10, blogger “November” said AI converted a point-reading package into another format and successfully performed real-device point reading on a Caterpillar Wi-Fi point-reading pen; within two weeks, a relay of Xiaokedou point-reading package conversions appeared on Xiaohongshu. The open-source project tolpo studies Tapo smart-home communication methods and connects lights, plugs, and sensors to AI Agents; Reddit user motojo, meanwhile, spent a month working with Claude to restore the use of old products from a sleep company that had gone out of business.
In the past, modifying devices required knowledge of Linux, Android, firmware, and programming protocols, and ordinary people often got stuck at a single line of error output. Today, Agents can look up information, analyze files, write programs, and continue checking and revising after execution errors. Users do not need to systematically learn the technology first; they can retrieve knowledge around the problem in front of them and complete the operation.
This change makes methods validated by one person easier to reuse. After the modification process is posted on Weibo or Xiaohongshu, others can continue asking questions with their own device models, needs, and error messages, turning onlookers into actual attempts. Under the Weibo topic #人工智能给电视去广告教程# (#AI tutorial for removing ads from TVs#), many authors explain the guides in detail, amplifying the public’s imagination of what AI can be used for.
For ordinary people, the AI dividend may first show up on devices they already own: the system, home screen, formats, and maintenance methods are no longer entirely set by manufacturers at the factory. For brands, concentrated discussion on social platforms can become product-demand feedback. A TV with a few fewer ads, a point-reading pen that can read one more set of books, an old device getting one more chance to work—all are perceptible changes.
