Key Notes
- The creator describes a prototype that connects Claude to force feedback while an ESP32 controller operates the local hardware.
- Reliability remains a development goal, so the demonstration does not establish that the device can open every lock.
- Repeatable testing would need to measure successes, failures and human intervention, as well as the contribution made by the AI model.
Claude has acquired an unusual side job: helping a homemade robot pick a lock. YouTube creator Dr. Kun Dániel has shown a motorized device that connects Anthropic’s AI assistant to an ESP32-C3 Mini controller, giving the chatbot a physical task beyond the screen.
In the October 24 demonstration, the creator describes a system that sends force feedback to Claude over Wi-Fi. The model interprets resistance from the lock’s pins and guides the process, while the microcontroller handles local control. The account does not identify which Claude model is used.
The distinction between those roles matters. The ESP32 is the connection to the motors and sensors; Claude supplies a separate layer of interpretation. This is an example of a language model being connected to machinery, rather than evidence that the model itself runs on a tiny electronics board.
The chip’s manufacturer, Espressif, lists the ESP32-C3 as a single-core RISC-V microcontroller running at up to 160 MHz, with 400 KB of SRAM and built-in Wi-Fi. Its datasheet describes a low-power controller suited to connected devices. The demanding language-model computation is distinct from that local hardware role.
A Prototype, With Reliability Still to Prove
Dániel says he wrote the custom code and has spent more than 200 hours on the project. The mechanism and frame are 3D printed. He plans to release the source files and documentation after further refinement, describing consistent performance as unfinished work.
That qualification puts the video’s promise in perspective. An edited demonstration can show that a machine performs a task under particular conditions. It cannot, by itself, establish how often it succeeds across unfamiliar locks, how much intervention it needs or whether an AI model improves the result compared with conventional control software.
A useful evaluation would separate those questions. The same hardware would need repeatable trials, with failures and human interventions counted alongside successful attempts. Comparing AI-guided operation against a fixed controller would help establish where the model contributes useful judgment. Those are engineering tests, rather than conclusions that follow from an attention-grabbing clip.
The broader idea is already appearing in other settings. Our coverage of lab equipment examined efforts to connect agents to scientific instruments. This maker project works at a much smaller scale, illustrating why the interface between a model and physical hardware deserves as much attention as the chatbot’s answers.
Software instructions become more consequential when a device can exert force. A practical system needs clearly bounded actions, reliable measurements and a way for its operator to stop it. For this prototype, the interesting question is whether Claude can turn sensor readings into useful decisions consistently enough to justify its place in the control loop.
Disclaimer: AIstify is an independent media brand owned and operated by NuvexMedia LLC, publishing news, research, and insights on artificial intelligence, emerging technologies, automation, and related industries. NuvexMedia LLC invests in and collaborates with companies across the AI, technology, software, and digital innovation sectors. These relationships do not influence AIstify’s editorial coverage, and the publication maintains full editorial independence to provide accurate, timely, and objective information. © 2026 NuvexMedia LLC. All rights reserved. This content is for informational purposes only and should not be considered legal, tax, investment, financial, or other professional advice.
