Guy Oseary of Sound Ventures: AI Venture Capital is Like the Flood of Demos from Years Ago
Guy Oseary, a general partner at Sound Ventures, compares the current AI startup market to the flood of demos he experienced when he entered the music industry at age seventeen: leaving a club, people would hand him demos, making it hard to decide which to sign.
He stated that with enough time and due diligence, the good and bad will gradually become apparent. At seventeen, facing hundreds of demos, the top three he initially favored would not even make the top twenty by the time he reached the hundredth demo. It requires extensive listening and meeting people. He directly compares the early work of discovering emerging artists to selecting founders to invest in. Oseary later managed artists like Madonna, U2, and the Red Hot Chili Peppers, and served as an executive at Maverick; in 2010, he co-founded A-Grade with Ashton Kutcher, and in 2015, co-founded Sound Ventures, with a portfolio that includes Airbnb, Uber, Spotify, OpenAI, and Anthropic.
The key point of the demo metaphor is not the "feeling" but that rankings can be overturned by sample size. The top three may hold in a small sample but fail in a hundred samples. This turns the AI funding frenzy into auditory fatigue: project narratives can be very similar, and only through continuous listening can rankings be rewritten. He previously likened founder pitches to albums and singles, stating that engineers are today's rock stars.
Sound Ventures has another $240 million fund focused on foundational AI, publicly emphasizing platform-level inflection points and narrative and distribution capabilities. Entertainment networks are used to increase exposure for consumer and brand companies, while model companies follow the same process of "recognizing the voice before expanding the audience."
In market mechanisms, this is a repricing of selection labor. The buyers are early funds overwhelmed by demos; the sellers are the emerging AI founders. The beneficiaries are those who can listen to the hundredth demo before placing an order; the pressured are those who close deals based solely on impressions from the top three. Funding still enters teams deemed "tour-ready" based on rounds, not on the number of demos themselves.
Supplementary structure: Oseary does not deny first impressions; he denies that first impressions remain valid as the sample size increases. The signing windows in the music industry and the seed windows in venture capital are, in his view, the same set of listening training.
Source: Public Information
ABAB AI Insight
The flood of demos at seventeen serves as Oseary's operational manual for AI venture capital in 2026: the top three can be overturned by the hundredth. The Madonna system trained him to find tour-worthy voices amid the noise; Sound Ventures applies the same listening to founders. The demos in the AI market are not tapes, but demo days and seed memos, where the scale has turned "feeling right" into a statistical error.
The capital path is time. The cost of listening to a hundred demos is ears and calendar, not valuation models. Those willing to spend their calendar on meetings and due diligence gain the right to correct errors after being misled by small samples. The $240 million foundational model fund indicates he acknowledges the next album is at the model level, yet the selection action remains A&R: first sorting people, then discussing rounds. Kutcher complements engineering intuition, while Oseary focuses on whether "this voice can be heard."
This parallels the independent label demo submissions of the 1970s and the flood of mobile app submissions post-2008: as entry becomes cheaper, the gatekeeper's role shifts from discovery to sorting. The industry phase is characterized by AI project supply exceeding early partners' listening bandwidth. Demo days cram hundreds of teams into the same week, increasing the mortality rate of first impressions.
Structural judgment shifts pricing power from the primacy effect to within-sample sorting. The mechanism is: when supply explodes, first impressions act as noise amplifiers; continuous listening turns noise into comparable rankings. The cost of signing the wrong demo is a failed album, while the cost of investing in the wrong founder is the opportunity cost of the entire fund. Thus, "listening more" is not just a platitude; it is the only skill that appreciates as sample size increases.
ABAB News · Cognitive Laws
The top three only hold when the sample is not large enough.
After supply explosion, the gatekeeper's job shifts from discovery to sorting.
Listening is the only due diligence tool that appreciates as projects increase.