Music startup Thoughtful Things has launched a Kickstarter campaign for its first instrument, Engram. It’s a sampler and groovebox that uses AI to mangle incoming audio and hallucinate completely new sounds. According to The Verge AI, the pitch is the opposite of what most AI music tools promise. Engram isn’t built to make polished tracks. It’s built to make strange ones.
That’s a different take on generative audio. Most of the attention in AI music has gone to tools like Suno, which turn a prompt into a finished, radio-ready song. The Verge AI puts it plainly: “This isn’t Suno in a box.” Engram treats AI’s glitches as the point.
🎛️ What Engram Actually Does
Here’s what makes this device stand out, based on the Kickstarter details reported by The Verge AI:
- It runs a “tiny AI” locally. Engram doesn’t connect to the internet. The model lives on the device and runs on its own hardware. That means no cloud subscription, no latency from remote servers, and no uploading your recordings to anyone’s servers.
- The model is custom-built and trained in-house. Thoughtful Things says it designed and trained its own audio models instead of licensing someone else’s. For a small hardware startup, that’s a big technical bet.
- The training data is licensed for commercial use. The company says: “We’ve trained our audio models on open datasets that only contain audio licensed for commercial use (CC-BY or similar). We have not trained and will never train our models on non-commercial, pirated, or otherwise stolen data.” That’s a pointed statement, given the lawsuits and debates over how music AI companies source their data.
- You can prompt it with your voice. In the campaign video, founder Evan King asks Engram for “piano.” What comes out is glitchy and only vaguely piano-like. That’s by design. King calls the device a “field recorder for latent space.” Latent space is the internal map of patterns an AI model learns, and Engram lets you poke around in it and record what you find.
- It borrows from circuit bending. Circuit bending is the old DIY practice of rewiring cheap electronics, like toy keyboards, to make unexpected sounds. Engram applies that spirit to AI. You can tweak its tiny models and even break them on purpose.
- The firmware will open up. Thoughtful Things plans to open Engram’s firmware so users can modify it or load their own custom models. If that happens, the device becomes a platform as well as an instrument.
💰 Price and Availability
Engram is launching as a limited Kickstarter run, with pledges starting at $675. The company hasn’t announced a retail price. The Verge AI notes that the $675 pledge is listed as a 30 percent discount, so the eventual retail price will probably land somewhere between $850 and $900.
The usual crowdfunding caveats apply. Kickstarter hardware projects often slip on shipping dates, and backers are funding a product that isn’t in stores yet. The open firmware is also still a plan, not a shipped feature.
🔍 Why This Matters
What stands out here is how Engram flips the AI music conversation. The industry has mostly chased realism: AI that sounds more human, more polished, more like the charts. Engram treats AI’s weirdness as creative raw material. The hallucinations people complain about in chatbots become the whole point.
That puts Engram closer to experimental gear than to a song generator. It’s for sound designers and electronic musicians who already love samplers and odd textures, not for people who want a finished track with one click.
The local, offline approach matters too. It sidesteps privacy worries and subscription fatigue, and it makes Engram feel like an actual instrument you own. Pair that with the clear stance on licensed training data and you’ve got a product that answers two of the biggest criticisms of AI music at the same time.
🎶 What Comes Next
The real test is whether a small on-device model can make sounds that musicians want to keep, not just novelty glitches. If Thoughtful Things follows through on open firmware, a community of modders could push Engram much further than the company could alone. It’s a niche product for now, but it hints at a future where AI shows up in music as an instrument rather than a replacement for the musician. You can find more details in The Verge AI’s original coverage.