An Algorithm…with a Record Deal?

How AI-powered start-up Endel is building a business creating music to help us relax

I am not going to lie — I really, really struggled to get started on this blog post. There are only two weeks left of school, the finish line is in sight, and yet here I am drowning in final projects and papers. I have honestly started to feel a bit mentally drained and recently found myself feeling some anxiety about how I’m going to get myself in the mental state to get everything done so I can enjoy the end of my time here at Harvard Business School. 

As someone who experiences a lot of joy from music, I began listening to my favorite artists on Spotify in the hopes that this would get me out of my funk. Though this improved my mood substantially, it certainly did not help with my focus (and actually led to more embarrassing solo dance parties). I then sought out playlists that claim to assist with focus, and in my research I ran into Endel. 

Endel is a mobile application that creates personalized, artificial intelligence-powered music based on an individual’s immediate conditions to reduce stress, improve sleep, and boost productivity. Most functional audio experiences (such as the focus-themed playlists on Spotify I was listening to earlier) are fragmented and non-adaptive. However, Endel produces music by actually drawing from live data from the cell phone of the user, including the user’s circadian rhythm, weather, and heart rate (via Apple Health or Google Fit). Using artificial intelligence, Endel then creates a custom music experience that promotes focus, mindfulness, and relaxation. Further, as the user’s input data changes, the music adapts in real-time to create an optimal experience for the user. For example, if Endel senses a prolonged increased spike in heart rate while the user is motionless (perhaps suggesting increasing levels of anxiety), the music adapts to help bring the user’s heart rate back to a resting level. 

So what does Endel’s music even sound like? Well — take a listen for yourself! Yeah…so not exactly Ariana Grande. But perhaps not surprisingly, this is intentional!

Upon first listen, you may note that the music is quite simple. Endel focuses on creating music of decreasing complexity to ease mental burden. Endel CEO Oleg Stavitsky explained that while complex song structures and unique melodies may sound nice, they force the brain to work a little harder to make sense of them. Therefore, for music composition, Endel uses a simpler pentatonic scale (in contrast with the heptatonic scale that most Western music uses), less complex tones, and simple melodies. Endel cites studies that suggest music is found to be pleasing when it triggers a rhythmically consistent pattern in certain auditory neurons, so their music further incorporates simple ratios of two tonal frequencies like octaves and perfect fifths. 

Endel further employs models to determine which sounds are best suited for relaxation and which are best suited for alertness and focus. Stavitsky cites studies that show that relaxation is best facilitated with mellow tones, slow chord changes, and simple structures. He says that our brains are constantly analyzing sound, and the less detail there is, the less attention is dedicated to that task. He believes this helps facilitate relaxation quicker and for longer periods. 

Endel’s technology is validated by the Experience Sampling Method by esteemed psychology scholar Mihaly Csikszentmihalyi, demonstrating that Endel increased concentration in users by 6.3x and decreased anxiety in users by 3.6x. 

Since Endel’s inception in 2018, close to 2 million users have joined the company’s ecosystem of consumer-facing products, helping Endel gain traction and land the accolade of App of the Day in the App Store in 138 countries in addition to being named one of the Best Apps of 2019 by Google. 

More recently, Endel has expanded its business model outside of the consumer applications. Endel became the first algorithm to sign a distribution deal with major music label, Warner’s Arts Music division. The deal was for Endel to use its algorithm to create 600 short tracks on 20 albums that were then put on streaming services, returning a 50/50 royalty split to Endel. Dimitry Evgrafov, Endel’s composer and head of sound design, says all 600 tracks were made “with a click of a button.” There was minimal human involvement outside of chopping up the audio and mastering it for streaming. Five Endel albums have already been released, and 15 more are coming this year — all of which will be generated by code. In the future, Evgrafov thinks that Endel will be able to make “infinite ambient tracks.”

Endel has further expanded its business model through partnerships. This past October, Endel announced that it had partnered with musical artist Grimes for a special collaboration called AI Lullaby, combining Endel’s artificial intelligence with Grimes’ original music and vocals (see above). Additionally, Endel has entered into a partnership with ANA Airlines, Japan’s biggest airline, to feature its technology in planes to enhance the comfort of passengers. 

A serious debate is beginning in response to Endel’s increased partnerships and collaborations. Unlike many AI-driven software packages for music creation (e.g. Google Magenta Studio, IBM Watson Music) that create tools for the musician to use, Endel is actually creating a finished product. Endel maintains that it is not competing with musicians, but some argue that the Endel/Warner deal in particular is a major inflection point, as the label made the decision to distribute and monetize automated audio, receiving 20 hours’ worth of materials for relatively little labor and expense. It’s hard to imagine that other labels won’t soon follow.

Endel’s goals are ambitious, and the company seems to be getting significant support, recently raising a Series A round. The company believes it is “democratizing wellness by making AI-powered wellness and mindfulness accessible to all,” though it will be interesting to see how this mission of producing AI-powered music co-exists alongside music produced by human artists. 


Carroll, Tobias. 2021. “AI Writes “New” Songs By Nirvana, Jimi Hendrix And Amy Winehouse”. Insidehook.

Deahl, Dani. 2021. “Warner Music Signed An Algorithm To A Record Deal — What Happens Next?”. The Verge.

Endel. 2021. “AI Startup Endel Invents Futuristic ‘Sound Environments’ That Reduce Stress, Improve Sleep And Boost Productivity”. Static.Endel.Io.

Ha, Anthony. 2021. “Endel Raises $5M To Create Personalized ‘Sound Environments’ That Improve Productivity And Sleep”. Techcrunch.Com.

Hayes, Tyler. 2021. “The Science Behind Endel’s AI-Powered Soundscapes”. Amazon Science.

Newman, Rebecca. 2021. “Lend Me Your Ears: Have You Heard Of Sound Wellness?”. ELLE.

Summers, N. 2021. “Grimes And Endel Bring Their AI-Powered Sleep Sounds To Android”. Engadget.Com.

Wang, Amy. 2021. “Warner Music Group Signs An Algorithm To A Record Deal”. Rolling Stone.


Darktrace: Cybersecurity of the Future?


Legacy Furniture Bohemoth IKEA, All In On AI

Student comments on An Algorithm…with a Record Deal?

  1. This is so fascinating! I feel like we often hear about how the role of computers and AI is relegated to science, so it’s really interesting to see how AI is being leveraged in art. I do think the value proposition is supported by the fact that the “art” in this case serves a very specific, and measurable, use case in creating calm. Who knows, one day maybe an AI can also compose music that is simply beautiful as well!

    1. Yes! I found this so fascinating as well. This will be a really interesting company to watch.

      I definitely read through a lot of tension felt by musicians and composers about what Endel is doing — and there is definitely debate about whether or not the music that Endel creates is truly “art.” I think its a fascinating debate.

      If you’re curious — I’d recommend this article:

  2. Holy cow, I had no idea this level of intuitive technology already existed. It’s crazy to think that this software can aggregate data from disparate sources stored on your phone and play appropriate music. I like where this technology is going in terms of it’s potential impact of improving the quality of people’s life. What does concern me is the creation of music without human input. This feels like a loss, not a gain for humanity. The implication of algorithmic music (or really any artistic) generation that concerns me is the disruption of a basic human cultural activity. I’m all for the ML algorithms matching music to mood, but protest ML creating art for us and hope that trend doesn’t continue. Thanks for sharing!

    1. Hey Omar — yeah crazy, right? I agree there are some potentially troubling implications here.

      If it is any consolation (at least given my understanding of the technology at play here), there are limits to AI-generated music. As we’ve learned in class, data in is data out, so though you could theoretically build a model on classical music, your model would create infinite variations of classical music, but wouldn’t be able to innovate to create jazz.

      So though Endel can now make infinite ambient tracks, it’s hard to imagine that the technology as it stands could create the next wave of musical innovation.

      I find comfort in this realization — because I think there (at least for now) is still room in art for human creativity and innovation that is uniquely human. My hope is that AI music tools can assist composers and writers, but certainly not replace human intuition and creativity.

  3. Pretty cool product, Daniel, thanks for sharing! What a cool mission statement: “democratizing wellness by making AI-powered wellness and mindfulness accessible to all.” On top of the Warner deal you wrote about, I wonder if this company could become an attractive acquisition for one of the leading competitive music streaming services (Apple Music, Amazon Music, Spotify) because of the uniquely personalized experience that it offers the user. I agree with Omar that, for better or for worse, this tech fundamentally changes one of the most basic human experiences.

    1. Hi Cristina — thanks for your comment! I completely agree — a fabulous mission statement, and it’s nice to see a company use AI in way that is intended for good!

      I totally agree with your point here — I especially think the technology is well-aligned Apple’s view of the world in particular, since the model building process happens entirely on the user’s devices and data is not sent to the cloud or sold to third-parties.

  4. I loved the very personal introduction, Daniel. I shared your sentiment, thank you for the openness! I have never heard about Endel but, after reading your post, I definitely need to give it a try. One of my reflections after reading this revolved around the input data points Endel is tracking: “user’s circadian rhythm, weather, and heart rate”. I wonder how effective those metrics are at assessing a person’s emotional state. While there are clear physical signs present, there are other mental and chemical factors that may also be influencing it. Do you know if Endel is planning to integrate other mechanisms (i.e. self-assessments) to better measure a person’s emotional state?

    1. Hi Juan Carlos — yes the end of the year can be tough! Definitely give Endel a try, I’ve actually been using it all week since I wrote this post and it’s been shockingly helpful.

      If you’re interested in the science behind the model — I’d recommend this interview, they go into quite a bit of detail and cite numerous studies around how they chose their data inputs:

      I didn’t stumble across any plans to integrate other mechanisms, but if not clear from the post, you do need to indicate what “mood” you’re going for — relaxation, focus, sleep, etc. — which I think helps. I would surprised if that wasn’t in the plans given that the model today actually doesn’t take in too many inputs as your comment suggests. I also think they are a bit limited to what is measured via wearables, so the advancement of wearable technologies will have a big influence the development of their models.

  5. This is just amazing. I’ve read the post while I listened to one of the Endel songs and that felt great. It’s so exciting to see those being generated by algorithms. With enough user feedback and so many data points, I imagine this becoming the best personal composer any person can have, at least for people who enjoy ambient songs.

    1. Hi Marcos — thanks for your comment! Yeah I’m pretty blown away by the app — I honestly have been using it all week and it really has helped me focus on getting all of these projects done. I’m excited to see where the company goes and how the technology evolves!

Leave a comment