Around 2009, Edward Balassanian found himself listening to electronic music that seemed in no hurry to finish. DJs were playing sets that lasted for hours. What caught his attention was their command of the journey: energy could rise, settle and gather again without the experience collapsing into a succession of separate songs. A software entrepreneur had encountered a performance with an unusually interesting interface. The audience was already part of the feedback.
He later called this discovery “long-form electronica.” A DJ could read the room and decide what should happen next. A recording had already made those decisions. Balassanian began thinking about the space between them. Could software arrange musical ingredients as the listener listened? Could the person pressing play have some say in the outcome? Aimi, the company he founded in 2019, grew around those questions.
There is a pleasing inconvenience in this proposition. The recorded song has an excellent feature: it arrives finished. You know when it starts, when it ends and when to attempt the chorus. Balassanian’s idea asks people to become comfortable with music whose arrangement can remain open. The tune has left a little work on the table. Some listeners may rather like the invitation.
Software with a sense of timing
Balassanian’s first professional education was in computing. He entered the University of Washington at 16 and graduated at 19. His degree was in computer science, earned with distinction. From 1990 to 1995, he worked at Microsoft. Long before he was asking music to respond to its audience, he was working inside the machinery that allowed computers to communicate.
He remembers an era when delivering an urgent network fix could involve boarding a plane with a floppy disk and installing it at the customer’s machine. That is a particular kind of customer contact. The engineer and the delivery mechanism occupied the same airline seat. A problem in software could still require somebody to carry a bag.
His subsequent businesses ranged across BeComm Corporation, which developed operating systems for digital devices, the incubator Be Labs, Vital Juice and the publishing platform Strings. That assortment can look eccentric on a business card. It also explains why he approaches music with a platform builder’s vocabulary. His career has repeatedly involved making systems through which other activities can happen.
“I consider myself a platform person. I build operating systems.”
Edward Balassanian · 2024 conversation
He is a named inventor on more than 80 patents, covering areas including gesture interfaces, networking, distributed processing and digital music synthesis. These interests meet at a practical question: how does a person direct a complicated system without having to manage every operation inside it? A volume control is easy. A control for the unfolding character of a performance is more demanding.
The DJ’s second job: teaching the founder
Entering music meant learning a different craft. Balassanian hired DJs and producers to teach him theory, mixing and the construction of music at the sample level. The detail matters because the original idea depended on more than a pleasant sequence of sounds. A set needs judgment about what belongs together and when a change should arrive. The expert can hear a problem before a novice knows what to call it.
When Aimi publicly launched its app in May 2020, it combined artificial intelligence with material supplied by electronic artists. The early experience generated music on the phone and allowed listeners to choose its energy. Balassanian described a distinction between making the musical fragments and arranging them into a performance. Software could take on some of that arrangement work while artists continued supplying the creative ingredients.
That division gave the company a concrete starting point. It could work with loops and decisions about how loops fit together. It could also ask listeners to respond to something while they enjoyed it, rather than assigning them the rather thankless occupation of evaluating bad music. Enjoyment had to come early enough to make participation worthwhile.
By 2023, Balassanian was describing an approach combining expert systems and machine learning. Production knowledge helped establish a listenable foundation; learning could then introduce preferences associated with a person, artist or genre. The explanation reveals his engineering temperament. He was interested in the steps that make a performance work, and in which of those steps could become adjustable.
The producer’s unfinished business
The artists gave this system its character. Aimi+ arrived in August 2021 with experiences from producers including Cassy, Shanti Celeste, Roska and Tensnake. Max Cooper and Soul Clap joined the following month. These were musical identities with their own habits and audiences. A system that flattened all those differences into one agreeable background would have lost much of the point.
The artists could provide beats, stems, individual instruments, scrapped ideas and unfinished songs. They configured creative parameters; the software rearranged the material into a continuous mix. An idea that had never made it onto a record could become useful again. The studio’s unfinished drawer had acquired an alternative distribution plan.
There was a social component, too. Balassanian hosted conversations with musicians in Aimi’s CEO Sessions series, including a 2021 discussion with Man Power. Hearing a producer talk about working with the system made the experiment less abstract. Somebody had to decide which sounds to bring, which constraints to set and whether the result still felt like their work.
In his 2023 conversation with John Koetsier, Balassanian resisted the prospect of building a machine primarily to imitate existing recordings. He put it bluntly: “The world has enough cheap music.” His stated interest lay in giving people more ways to participate. An interactive player could let a newcomer change musical elements without first acquiring a producer’s vocabulary. Turning a control could precede knowing its technical name.
The price of another possibility
In November 2021, Aimi raised a $20 million Series B led by Great Mountain Partners, with Founders Fund participating. The company’s announced plans included desktop tools for creators and an updated operating system with a programming language. The investment connected a listening experiment to the more expensive ambition of building infrastructure around it.
Led by Great Mountain Partners
with participation from Founders Fund
Balassanian has also written about the less photogenic business of raising money. His advice to founders includes finding investors prepared to stay, leaving enough time for the process and being honest with employees about the position of the company. He cautions against presenting a rosy picture until the moment payroll becomes a problem. Those prescriptions put a practical limit on the romance of building something new.
The business itself changed emphasis. In an interview recorded in July 2024 and released the following March, he discussed enterprise customers, licensed content and reducing the team after reaching what he described as product-market fit. He regarded that reduction as beneficial. A useful application for a technology can narrow the organization required to deliver it.
Business customers brought another consideration into the foreground: permission to use the music. A soundtrack for a game, a video or a commercial environment has obligations beyond sounding appropriate. Balassanian discussed reducing that uncertainty as part of Aimi’s offer. For a founder interested in platforms, the rights attached to an ingredient become part of the product’s construction.
Leave the faders within reach
The current Aimi website centers on Session, a music studio in the browser. Its AI producer creates editable projects with tracks, effects and arrangements. People can direct changes conversationally or use the controls themselves. MIDI and virtual instruments are part of the offering. The project remains available for revision after its first generation.
This is a different encounter from the early adaptive player, yet the connection is audible. Both leave room for a person to intervene. One listener might want to influence the flow without becoming a musician. Another might want to adjust a particular part until it sounds right. A tool has to make those choices available at a level the person can use.
The distinction becomes ordinary when you imagine disliking a bridge. Regenerating an entire track is one response. Reaching into the arrangement and changing that section is another. Balassanian’s longstanding interest in interactive music gives the second response a history. The user has somewhere to put an objection. Even the bassline can receive editorial notes.
What the person still brings
In a January 2024 essay, Balassanian argued that wider access to knowledge through AI would increase the importance of applying it creatively. His position places value on judgment, expression and original ideas. In music software, that belief creates a fairly specific obligation: leave the person with meaningful decisions to make.
There is tension inside that ambition. Making a tool accessible requires the software to handle complexity. Preserving agency requires it to expose enough of that complexity for a person to act. Every new control adds another decision; every simplification removes one. His products offer different attempts to find a workable balance, with the listener or creator close enough to hear the consequences.
The thread running from those long DJ sets to an editable browser session is an interest in what happens next. A performer listens to the room. A listener changes the energy. A creator revises a passage. Balassanian keeps making software around the possibility that the next choice should remain available. The music can carry on. Someone still gets to touch the fader.