A startup run out of a vinyl bar in Tokyo's Shibuya district is betting that the next wave of music apps won't be built around typing prompts into a chat box. According to TechCrunch, the company — founded by a former Spotify executive — is releasing experimental "singles" that pull users directly into the act of making music. The location is not incidental. A vinyl bar is a room built for listening together, for handling physical records, for the small social rituals that streaming quietly erased. That the company chose this as its home base tells you most of what you need to know about its product philosophy.
What Is the Vinyl Bar Shibuya Startup?
Shibuya has long functioned as a laboratory for Japanese music culture, from the stacked listening booths of Tower Records to the tiny basement venues where entire subgenres gestate. A vinyl bar sits at the quieter end of that ecosystem: a counter, a turntable, a curated shelf, and a bartender who reads the room. It is a business model built entirely on human judgment about what a specific group of people might want to hear next.
The startup that TechCrunch reported on takes that setting and turns it into a product studio. Its founding team includes a former Spotify executive, which matters more than it might first appear. Spotify's core asset has never really been its catalog — it's the behavioral data generated by hundreds of millions of listeners: skip rates, playlist sequencing, repeat behavior, the precise second at which a song loses someone's attention. An executive who spent years inside that machinery understands better than most where algorithmic recommendation succeeds and where it produces sameness.
The company's output, per the report, is a series of experimental "singles." That word carries weight. A single is the traditional unit of a musical release — one song, delivered to the public, complete. Reusing the term for an app suggests the company thinks of each product as a discrete creative work rather than an incrementally updated platform. In an industry where apps ship constantly and pivot endlessly, releasing apps the way a band releases singles is a deliberate posture.
Music Apps Without AI: A Deliberate Choice
The global recorded music industry generated roughly $29.6 billion in trade revenue in 2024, according to IFPI's annual Global Music Report, with streaming accounting for about two-thirds of that total. Meanwhile, MIDiA Research has tracked a steady rise in tools that generate music from text prompts, and the volume of AI-produced tracks uploaded to streaming platforms has grown sharply enough that distributors have publicly acknowledged the strain on moderation systems. Deezer reported in 2025 that a substantial share of daily uploads to its platform were AI-generated.
Read next Top Technology Trends in 2026 You Need to KnowAgainst that backdrop, building a music app explicitly without AI prompting is a position, not an oversight. The TechCrunch report frames the absence of prompting as the point. Prompting reduces music-making to a text interface — describe what you want, receive a finished file. It is fast, it is accessible, and it removes nearly every decision that makes creative work feel like yours.
For a former streaming executive, the calculus is legible. Recommendation engines trained on skip data are excellent at serving listeners something acceptable and mediocre at producing surprise. Prompt-based generation is excellent at producing volume and mediocre at producing attachment. Both optimize for throughput. The vinyl bar startup appears to be optimizing for the opposite: fewer outputs, more involvement, and a result the user actually feels ownership over.
How the Experimental Singles Format Works
TechCrunch describes the releases as experimental singles that involve users in music making. The specific mechanics beyond that are not detailed in the reported summary, so the honest read is that the company is still defining the format in public.
What the framing does establish is a cadence and a unit of measure. A single is finite. It has a release moment. It can be anticipated, discussed, and then superseded by the next one. That structure runs counter to how most music apps behave — as always-on services engineered for retention metrics and daily active users. Releasing "singles" implies the company is willing to let individual products live and die on their own merits, closer to how a label operates than how a SaaS company does.
It also implies experimentation without permanence. A traditional app release carries the weight of a roadmap; every feature is a commitment to maintain. A single can be weird. It can be narrow. It can serve one specific interaction and then stop. That latitude is rare in consumer software, and it's the kind of thing that tends to produce genuinely novel formats rather than incremental refinements of existing ones.
Involving Users in the Music-Making Process
The participatory instinct here has precedent, though it has rarely been well served by commercial software. Nineties CD-ROMs let listeners remix tracks from major artists. Björk's 2011 Biophilia shipped as an app suite with interactive modules built around each song. More recently, platforms like Splice turned sample libraries into a collaborative substrate, and Roblox and Fortnite have hosted in-game concerts where the audience's presence shapes the event. At SXSW and across Music Ally's conference coverage, a recurring theme among music-tech founders has been that passive consumption has hit a ceiling and that the next growth area is some form of active participation — karaoke, remixing, stems, social listening rooms.
What has been missing is a format that involves users without demanding that they become producers. Full DAWs are intimidating. Prompt boxes are frictionless but anonymous. The middle ground — structured participation where a user contributes something meaningful inside guardrails someone else designed — is where most of the interesting work is happening, and it's where the vinyl bar startup appears to be positioning itself.
Analysts covering music tech have made a related point for years: the streaming era unbundled the album and then failed to rebundle anything in its place. Playlists are not scenes. Algorithms are not communities. A vinyl bar, notably, is a scene — a physical place where taste is transmitted socially. Translating that into software is hard, but it's a more credible thesis than another recommendation engine.
The Broader Shift Toward Human-Centered Music Tech
The timing is worth noting. Vinyl sales have grown for eighteen consecutive years, per RIAA data, reaching levels not seen since the early 1990s. Cassette reissues sell out. Physical formats now function as statements of intent rather than distribution necessities. People are not buying records because they sound better in any measurable sense; they're buying them because the act of choosing, handling, and playing a record is a form of participation that streaming doesn't offer.
That same appetite shows up elsewhere. Vinyl bars have opened in London, Brooklyn, Mexico City, and Seoul. Listening clubs and ambient listening parties have proliferated. The common thread is deliberate friction — a small amount of effort that makes the experience feel chosen rather than served.
A startup operating from a vinyl bar in Shibuya and shipping non-AI music apps sits squarely inside that trend. It's not anti-technology. It's a bet that the industry's most valuable resource is no longer access to music, which is effectively infinite, but attention and attachment, which are not.
What This Means for the Future of Music Apps
The near-term question is whether a company built on experimental singles can sustain a business. Experimental formats often generate attention and rarely generate revenue at scale, and releasing apps like singles means accepting that most will be discarded. That's a hard sell to investors accustomed to recurring revenue, though it's a familiar model to anyone who has worked in music itself.
The more durable significance may be the signal it sends. An executive with deep visibility into streaming behavior chose to leave that world and build something that rejects prompts, rejects infinite catalogs, and rejects the assumption that more automation is always better. If the format finds an audience, it will be because a meaningful number of listeners want to do more than press play — and are willing to accept a bit of friction in exchange for feeling like the music is partly theirs.
Source: TechCrunch
