questions, answered.

a continuous music stream you describe in plain language. type a moment (a scene, an activity, a feeling, anything specific to where you are right now) and the music starts. it keeps going. when you want something different, you type a new moment and the stream shifts.

the closest existing reference is the lofigirl youtube livestream, except it isn't just lofi and it's tuned to whatever you type.

no. suno makes one song at a time for creators to iterate on, save, and release. flowy makes a continuous stream for listeners. there's no save button on a single track, no lyric editor, no “extend by 30 seconds”. the whole product is the stream and your description of where you are.

we're in a different category. if you want a song to keep, go to suno. if you want music in the background of a moment, that's flowy.

there's a free tier. anonymous visitors get three plays per station and two stations per browser. signed-in free users get up to 90 minutes of streaming per day and can save tracks to their account. unlimited streaming + every mp3 download free is on the subscription tier.

the subscription is intentionally priced below comparable services. listening time costs real money per second, so a forever-unlimited free tier isn't feasible. the free tier is generous enough to actually use.

between €0.50 and €1.99 depending on the price test you're in. subscribers get every download included.

yes. flowy is operated under eu consumer-rights law: any digital-goods purchase (mp3 downloads, subscriptions) carries a 14-day right of withdrawal under § 356 bgb. for downloads, the right of withdrawal expires once the file has been delivered (you're asked to confirm that at checkout). subscriptions can be cancelled any time from the stripe billing portal.

if something feels wrong, email me: contact.timgrossmann@gmail.com.

short answer: write a scenario, not a genre. “afro house” on its own gives the model the average of every afro house track in its training. “afro house warmup for guests arriving on a saturday” gives it something specific to anchor to, and the music lands closer to what you actually wanted.

this is a real issue and a real fix is coming. a russian dj flagged it last week with afrohouse specifically, and he's right. the model occasionally drifts to neighboring styles on pure-genre prompts. scenarios constrain it. on the roadmap is better fidelity for pure-genre prompts too, but for now: describe the moment, not just the style.

half on purpose, half a thing to fix. the stream stays in the style you described, so a “deep focus binaural background” moment is supposed to feel coherent across tracks. but if you keep typing the same kind of moment three days in a row, the outputs do start to live in the same sonic neighborhood.

two fixes. for variety inside one session: tracks within a stream are deliberately varied even though they share the mood. for variety across sessions: rotate your moments slightly. “classical strings late night essay” one day, “japanese city pop neon tokyo” the next. same emotional register, different sound entirely.

honest answer: that's a real bug that's been reported by a handful of early users and it's the single thing i care about fixing first. if it happens to you, please email me with what you were doing right before it broke. browser, device, whether you switched tabs, any detail helps.

a recommendation: signed-in users who hit this can usually resume by typing the same moment again. the stream picks up fresh tracks from where the last one stopped.

short answer: the legal ground under ai music is still actively moving. flowy's position is that mp3 files you've paid to download are licensed to you for personal and non-commercial use. for commercial use (monetized youtube content, advertising, paid client work), email me and we'll work it out. the answer is usually yes, we'll just need to put something in writing.

this is going to harden into a proper license over the next few months. for now, ask.

legally murky everywhere in 2026. most jurisdictions hold that purely ai-generated output isn't copyrightable by anyone (no human author = no copyright). flowy doesn't claim copyright on what gets generated from your moment. you're paying for the convenience of the stream and the file delivery, not for an exclusive license to a piece of original authored music. consult a lawyer if your use case depends on enforceability.

within a single session you're streaming, no. across sessions, signed-in users get a stronger guarantee: tracks you've played in any prior session never play again. anonymous browsers don't carry that memory across reloads, so the same cached track may surface twice.

the moment input accepts any language. internally, the model is comfortable with all major european languages plus most widely-spoken non-european ones (japanese, korean, mandarin, hindi, arabic). the ui is in english by default, and there's a full russian version at flowy.fm/ru. more locales are coming.

usually yes. a moment in spanish produces music with spanish vocals where vocals make sense. a moment in hindi produces bollywood-style instrumentation. when you write an english moment about a non-english scene (“tokyo neon”, “lagos rooftop”), the system reads the cultural cue and biases the music there even though the prompt language is english.

stripe doesn't accept russian cards today, which is a problem i want to fix soon. for now: foreign card, cryptocurrency, or just stay on the free tier (3 plays anonymous, 90 minutes a day signed in).

native crypto checkout (usdt and a couple of others) is on the roadmap and high priority. roughly a third of early flowy traffic is from russia.

if your moment has a near-match in the cached corpus (most common scenes do), the first track plays in well under a second. if it's a totally novel moment, you're waiting for fresh generation, typically 30 to 60 seconds. during that wait the player shows a warmup state rather than silently spinning.

not a native app. flowy works as a progressive web app. on ios safari, tap share, then “add to home screen” and you get a full-screen icon that behaves like an app. lock-screen controls and cover art work. on android, chrome will offer to install it directly. native apps may come later but the pwa shape is good enough for now.

the stream needs a network connection, so a full offline mode isn't possible. mp3 downloads work fully offline once you have the file. downloaded tracks live in your account so you can re-download from any device.

email (when you sign in), the moments you've typed, the tracks you've played / liked / saved / skipped, your purchase history (via stripe), and standard analytics events gated behind the consent banner. no ad tracking, no third-party pixels. data is hosted in the eu (frankfurt). full breakdown is in the privacy modal (the small dots in the bottom-right of any page).

any time. open the pricing page, sign in, click “manage subscription”. you land in the stripe billing portal with a cancel button. your subscription stays active until the end of the current billing period.

spotify's focus playlists are great until you've heard them. their catalog is finite, so within a few weeks of daily use you start noticing the loop. brain.fm is engineered focus music with a neuroscience pitch, which is great for focus specifically but doesn't cover the rest of your day. flowy generates fresh tracks every time, doesn't loop, and works for any moment you can describe. more on this on the vs spotify and vs brain.fm pages.

contact.timgrossmann@gmail.com. i actually read these. a one-line “this thing was confusing” is more useful than you think.