Music Startup Thoughtful Things Launches Engram, an AI Sampler for Uncanny Sound Design

Music technology startup Thoughtful Things has introduced its debut hardware instrument, Engram, via a crowdfunding campaign on Kickstarter. As reported by The Verge, the standalone groovebox and sampler employs artificial intelligence to manipulate incoming signals and generate unpredictable audio output based on machine learning hallucinations.

Rather than aiming to automate full track creation for commercial radio formats, Engram focuses on experimental sound design and unconventional sonic textures. Founder Evan King has described the device as a "field recorder for latent space," taking visual and functional inspiration from circuit bending. The unit encourages musicians to push its lightweight artificial intelligence models past standard operation limits. During a demonstration, King requested a "piano" tone via voice command, prompting the device to generate a glitchy, acoustic-style sound fragment.

Engram operates completely offline without requiring an active internet connection. The hardware executes an in-house, custom-trained compact model locally. According to Thoughtful Things, the underlying algorithms were trained exclusively on open datasets consisting of commercially licensed material, such as Creative Commons CC-BY audio. The company stated that it refrains from using unauthorized or copyrighted media. Furthermore, Thoughtful Things plans to make Engram's firmware open source, giving users the ability to modify the internal software or upload custom audio models directly to the hardware.

The initial production run of the instrument is being offered to early backers on Kickstarter starting at $675. That tier reflects a 30 percent discount, indicating that the final retail cost will likely range between $850 and $900 upon broader market availability.

Why It Matters

Engram highlights a strategic shift in how generative technology can be integrated into specialized musical instruments. While mainstream consumer software typically leverages cloud connectivity and broad web data to automate composition, Thoughtful Things is taking the opposite route: offline hardware processing, open-source firmware, and strict adherence to ethically sourced training data. This approach addresses lingering industry concerns surrounding copyright usage and internet dependency in hardware performance.

Additionally, treating artificial intelligence imperfections and hallucinations as intentional sonic features rather than technical defects reinterprets generative models as raw instruments for experimental art. By offering local execution and custom model loading, Engram bridges the gap between traditional tactile sound manipulation and modern machine learning, catering to electronic musicians seeking unique textures rather than automated song generation.