There is a Spotify playlist running in the cafe down the street from your office, and there is one running in the cafe at the other end of town, and somehow they are exactly the same. Different beans, different chairs, different clientele, identical sonic wallpaper.

That's the inefficiency Tonada is trying to attack. The Stockholm startup stepped out of stealth on Tuesday with backing from Vercel CEO Guillermo Rauch, a Last.fm co-founder, and a roster of music-industry operators. The product: AI-generated background music for retail and hospitality spaces, designed to actually fit the room.

Tonada was founded by Juan Manuel Serruya, previously an engineering lead at Spotify, and Jonathan Andersson, formerly head of sales at Wolt. The pair are pitching what they call composable, generative background music, tuned to time of day, foot traffic, occasion and brand. Tech.eu first reported the launch.

Background music in physical spaces is a strange industry. It's worth billions globally. It's almost entirely invisible. A small handful of B2B music providers, names like Mood Media, Soundtrack Your Brand, and Soundsuit, license catalog tracks to retailers and venues. The tracks come pre-cleared for commercial use. The retailer hits play. The shoppers buy more, supposedly.

Tonada's argument is that this whole stack is built on top of a fiction. The cafe down the street isn't actually playing music chosen for its space. It's playing a generic playlist licensed by its parent company three years ago, set to shuffle, looped forever. The retailer has no good way to test alternatives. The music provider has no real incentive to offer them.

There's a research case underneath this pitch. Tonada points, per its launch coverage, to the well-documented finding that the tempo of background music can move retail sales by more than 30 percent. Slow tempo, customers linger and spend more on browsing categories. Fast tempo, throughput goes up at lunchtime. Most retailers leave that lever entirely untouched.

If that's the opportunity, the question is whether anybody's actually willing to pay to capture it.

Here's where things get interesting, and a bit murky. Generative music is having its commercial moment. Suno and Udio are getting sued by major labels. Stability has its own audio model. Meta has trained one. The legal question of who owns the output, and whether the training data was lawful, is unsettled in every market that matters.

Tonada is pitching to retailers and hospitality operators. Those are commercial users. Commercial users care about license risk more than consumers do. A coffee chain doesn't want a letter from a label four years from now arguing that the AI track playing in store 47 was trained on copyrighted material.

The company hasn't fully detailed its training data approach in public materials. The investor list, though, includes a Last.fm co-founder. That's an interesting tell. Last.fm spent a decade negotiating between technology and labels. Whoever's involved knows where the bodies are buried.

Round size hasn't been disclosed. The investor profile reads like a seed bridge or large pre-seed, with strategic value weighted heavier than capital. Rauch in particular tends to angel into companies he can imagine on Vercel's deployment graph, which gives Tonada a quiet edge in any future enterprise sales motion.

Three things that could go sideways.

First, retail music is a cost center, not a revenue line. Most physical retailers think about their music budget in the same bucket as their lightbulbs. They want it cheap, reliable, and invisible. Convincing them that AI-tuned, brand-aligned, foot-traffic-aware audio is worth a meaningful premium is a long sales cycle.

Second, the incumbents will copy fast. Mood Media has thousands of enterprise contracts. The moment AI-generated background music shows real revenue traction, Mood will ship its own version. Soundtrack Your Brand has Spotify in its DNA. Soundsuit is European and well-positioned. The first-mover window in this category is shorter than it looks.

Third, the cultural pushback is real. Working musicians, songwriter unions, and label adjacent groups have started flagging AI-generated commercial music as a category they want regulated out of existence. Sweden, where Tonada is based, is one of the most pro-creator licensing regimes in Europe. That cuts both ways.

The bull case is just as real. If retailers genuinely care about their physical brand experience, and the data backs them up that ambient music drives behavior, then a tool that tunes that music dynamically to space, time, and customer is a clear upgrade over a static playlist. The question is whether that upgrade is worth a five-figure annual contract per location, which is what the unit economics need to work.

There's a not-quite-coincidence in Tonada being a Swedish company. Sweden produces a wildly disproportionate amount of the world's commercial pop. It also produced Spotify, which redrew streaming for consumers. The B2B music infrastructure layer has been the next obvious shoe to drop. It just took an AI model to make it interesting again.

Serruya's Spotify background is doing real work here. The hard part of selling music to physical spaces isn't the audio. It's the metadata. Knowing which song works in which moment is a question Spotify engineers have been quietly answering for fifteen years. Tonada gets to skip several years of that learning curve.

Andersson's Wolt experience is just as load-bearing. Wolt scaled into hundreds of cities partly because its sales team learned to pitch tiny, conservative restaurant operators on technology they didn't ask for. Background music providers have an even harder version of that sales motion. It's not obvious there's a better team for it.

The licensing approach is the first thing to watch. If Tonada releases public terms confirming that its model is trained on licensed catalog plus internally-generated material, the legal risk shrinks dramatically. If it stays vague, expect questions from labels within a year.

The second is the launch customer roster. Retail and hospitality is segmented enough that getting one anchor brand in apparel, one in coffee, and one in hotels would tell us the product travels. A long tail of independent boutiques would suggest it's a self-serve play, which is a different business with different economics.

The third is whether Tonada starts to publish results. Background music vendors love claiming sales lift. They almost never publish the data. A startup willing to A/B test against a control group, and share the numbers, would do something the category has resisted for thirty years. That's the kind of move that defines a category leader. Or, occasionally, gets one humbled.

Either way, this is one of the more interesting B2B AI launches of the spring. Most generative AI companies in 2026 are pitching at consumers or developers. Tonada's betting on the strange, sleepy, lucrative middle layer where physical spaces still run on intuition. That's a contrarian bet. Those tend to be either great or instructive.

Worth flagging the elephant. Spotify already runs a B2B background music service, Soundtrack Your Brand, which it spun out years ago and now licenses through. The product is decent. The catalog is huge. The pricing is reasonable. Most retailers who think about background music as a line item buy from Soundtrack or Mood and call it done.

Tonada's pitch has to clear that hurdle. The advantage of generative audio is that it can be tuned to a specific room, a specific brand, a specific time of day, in a way no fixed catalog can match. The disadvantage is that retailers have to be persuaded the upgrade is worth a higher price. The middle of the market won't pay it. The high end might.

Hospitality is potentially a stronger entry point than retail. Boutique hotels, restaurant groups, and gyms care about brand atmosphere in a way most fashion retailers stopped caring about a decade ago. They also have less standardized music procurement, which means a startup can land deals one operator at a time without having to win a corporate-wide rip-and-replace.

That's probably where Tonada lands first. Watch the early customer logos for confirmation.