# Engram: An AI-powered experimental music sampler A new instrument transforms AI hallucinations into glitchy, uncanny sounds. - **Source:** [The Verge](https://www.theverge.com) - **Published:** 2026-09-27 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/8f73ee56-9797-4b79-bc08-c973d09cc82b ## AI-powered experimental instrument Engram is a new sampler and groovebox from Thoughtful Things that uses AI to manipulate audio and generate experimental sounds. Unlike mainstream AI music tools, it focuses on creating uncanny, glitchy outputs rather than polished tracks. The device runs a custom-trained 'tiny AI' model locally, without internet connectivity, and draws inspiration from circuit bending to push AI audio models beyond their limits. ## Latent space exploration Described as a 'field recorder for latent space,' Engram allows users to explore the boundaries of AI hallucinations in sound design. Users can input sounds like voice commands and receive distorted, experimental outputs. The device encourages creative manipulation of its AI models, even letting users 'break' them intentionally for unique sonic results, much like traditional circuit bending techniques with hardware. ## Ethical training approach Thoughtful Things emphasizes its ethical approach to AI training, using only open datasets with commercial-use licenses (CC-BY or similar). The company explicitly states it avoids non-commercial, pirated, or stolen data. This positioning comes amid growing industry concerns about AI training data sources, particularly in creative fields like music where copyright issues are prevalent. ## Open firmware strategy The company plans to open Engram's firmware, allowing users to modify it or load custom AI models. This open approach contrasts with many proprietary AI music tools and aligns with experimental music hardware traditions. It enables a community-driven development path for the instrument's capabilities, potentially leading to unexpected sound design applications beyond the creators' original vision. ## Pricing and availability Engram launches via Kickstarter with early bird pricing starting at $675 (a 30% discount). The expected retail price ranges between $850-$900. This positions it as a premium experimental instrument, comparable to high-end boutique music hardware. The initial limited production run suggests Thoughtful Things is targeting professional sound designers and experimental musicians first. ## Market differentiation Engram deliberately avoids the 'push-button, get-song' approach of AI music tools like Suno. Instead, it focuses on the creative potential of AI's imperfections and hallucinations. This niche positioning helps it stand out in the crowded AI music space while appealing to experimental artists who value unpredictability and unique sound textures over commercial polish. ## Technical implementation The device runs a custom, in-house developed AI model optimized for local processing without cloud dependencies. This technical choice enables real-time manipulation and performance use cases where internet latency would be problematic. The local processing also addresses privacy concerns and allows for more radical sound experimentation without corporate content restrictions. ## Future implications Engram represents a growing trend of specialized AI instruments that embrace rather than hide AI's quirks. Its success could inspire more tools exploring AI's creative potential through its failures and hallucinations. The open firmware approach may also set a precedent for community-driven AI model development in music technology, challenging closed proprietary systems. ## Sources covering this story - https://www.theverge.com/ai-artificial-intelligence/1001193/engram-sampler-ai-hallucinations-music — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._