Discover
Recoup Podcast
Recoup Podcast
Author: Patrick Sweetman
Subscribed: 30Played: 1,046Subscribe
Share
© Patrick Sweetman
Description
Recoup Podcast is a show about AI transformation for music rightsholders. We sit down with the people who create, own, and invest in music, from label founders and publishers to fund managers and catalog owners, on how they run their businesses and where AI now does real work inside them. Specific answers from operators, the numbers behind them, and what they would build next. Made by Recoup, the firm that builds AI systems music companies own and trains their teams to use them. recoupable.dev
430 Episodes
Reverse
Paula Moore spent thirty years in A&R, built Warner's research platform Flypaper, then spent "a year of 16 hour days" building CreativeONE: a studio where every artist gets 100 specialized AI coworkers, with humans still making the calls.Paula opens the studio live. We meet Velar, the release strategy specialist, and the general manager that hands work to the rest of the team, so an artist never has to manage "100 new relationships" at once. Then the catalog side: a data room that replaces the usual Dropbox folder of spreadsheets, and five years of statements from a 20-year catalog, organized for a buyer.We also cover what her scouts read that a streaming chart misses, how she tells a real fan base from a one-song spike, whether she would license a catalog to Suno ("it's not truly my decision"), and the AI coworker she would hire first.Paula Moore: greatermusicgroup.com · creativeone.world · LinkedInChapters0:00 Cold open + intro0:16 From scout to builder4:48 A year of 16-hour days9:20 Coworkers, not tools19:18 The catalog studio26:59 Scouting past the spike28:26 AI music and Suno31:21 Quickfire33:24 OutroRecoup designs and builds the systems behind conversations like this one: catalog data, royalty operations, and AI agents for labels, publishers, funds and management companies. recoupable.dev
Zac Kubilus produced Frank Foster's Rhythm and Whiskey, which reached #4 on Billboard's Top Country Albums in 2014. In 2016 he recorded a $15 box fan from Walmart and put it online. Today his top tracks are box fans, and Sleep Sounds is his full-time work.Zac owns his recordings, his rights and his management, so he can try a new channel as soon as it opens. When Alexa arrived, Sleep Sounds became an Alexa skill. He explains why one owner now releases under three artist names after Amazon called "Sleep Sounds" too generic, and why the rules for functional audio now change in weeks or days instead of years.We also cover whether AI companies should train on a catalog (his answer: it's up to the owner), whether an owner should train a model on recordings they own, Nashville writers demoing with Suno, whether AI output needs a label, and what selling direct has taught him that streaming didn't.Zac Kubilus: sleepsounds.org · sleepmusicapp.com · Sleepy Buddy on Amazon Music · [email protected]:00 Cold open + intro0:20 From a #4 album to box fans5:38 One owner, many names8:22 The new rules for functional audio10:30 Catalogs as AI training data13:10 A model trained on your own recordings17:29 Selling direct21:40 Quickfire23:40 OutroRecoup designs and builds the systems behind conversations like this one: catalog data, royalty operations, and AI agents for labels, publishers, funds and management companies. recoupable.dev
"The lower multiple you pay, the more room you have to be wrong," says Tom Sarig, founder of AntiFragile Equity Partners, which buys catalogs from artists earning $100,000 to $1 million a year. He aims for 5x to 10x, runs a valuation in-house with his CFO, and must get a third-party valuation before he can draw on his credit line.Tom managed Lou Reed, and he is precise about where that money came from: Walk on the Wild Side and Perfect Day earned about $1.5 million a year, "100% sync." From his label years he tells how Cannons' "Fire For You" went into Netflix's Never Have I Ever for "like $3,000 a side" and, in his telling, went from about half a million streams to 50 or 60 million.Then the plan for Hot Vibes, the European dance catalog his firm bought: about 180 songs earning close to $400,000 a year, 19 ideas to grow it (SoundExchange registration, Musicbed micro sync, a sync rep, paid social), a hoped-for 10 to 15% a year, and AI remixes that could take it to 500 songs by the end of next year, as long as the melody and lyrics are "not composed by a computer."We also cover why he buys the whole album where, as he sees it, Duetti buys the one biggest track, how he screens for bot streams (four or five years of history, nothing declining more than about 5%), and whether Duetti's index of 3.8x to 10.2x matches what he pays: "That's about right."Tom Sarig: antifragilemusic.com/equitypartners · LinkedInChapters0:00 Cold open + intro0:24 Your purchase price is everything5:07 Two songs, a million and a half a year11:19 Finance people vs music people14:47 Total serendipity19:42 180 songs, 19 ideas22:37 The AI stack25:31 Bots, fraud and real growth26:52 What multiples look like now27:17 Quickfire28:21 OutroRecoup designs and builds the systems behind conversations like this one: catalog data, royalty operations, and AI agents for labels, publishers, funds and management companies. recoupable.dev
"Nobody actually wants a marketing plan," says Ali Lieberman, VP of Product at Symphonic Distribution. "They want the release to do well." That idea shaped the Release Campaign Builder, which Symphonic launched in February: a marketing plan built from a release's own data. Ali walks through one on screen, a demo release with a $500 budget and one goal, a playlist placement. One of Symphonic's senior A&R leaders called the output "annoyingly good."Before Symphonic, Ali ran the US ISRC agency at the RIAA and improved business processes at SoundExchange. Her rule for AI comes from that work: it is "mostly an amplifier." Clean data and processes get faster; messy ones get you "amplifying mess." She explains which work Symphonic has handed to AI, how the team defines and tests those workflows, and what stays with people.We also cover distributors as gatekeepers after UMG sued DistroKid and IFPI launched its Streaming Integrity Initiative, why the problem is "music pretending to be something that it isn't," why the default for AI training "should be no," and whether a model can price an advance: "I do not trust AI 100% to give the final number."Ali Lieberman: symphonic.com · LinkedInChapters0:00 Cold open + intro0:16 From ISRCs to AI3:40 Your super assistant8:36 A release plan written by AI13:35 Distributors as gatekeepers17:42 Opt in, not scraped22:00 Can a model price an advance?26:06 QuickfireRecoup designs and builds the systems behind conversations like this one: catalog data, royalty operations, and AI agents for labels, publishers, funds and management companies. recoupable.dev
"Our target multiple is a 20X," says Sean Peace, who founded SongVest in 2007. A catalog owner buys at 8x, then sells 10% or 15% of a song to fans at 20x. Fans pay it because a SongShare is "the coolest piece of memorabilia you'll ever buy": a piece of a song they love that actually pays a royalty.We test that on one real offering. In SongVest's 2026 offering circular, a share of "Rain On Me" earned $12,637 over the last twelve months. The seller gets $252,750, about 20 times that, and fans pay up to $318,465, about 25 times. Sean explains who those buyers are, how payouts reach them, and why the first song he put on the marketplace, "Hit The Quan", was one he owned himself.We also cover fewer, bigger buyers after Sony and Recognition and BMG and Concord, what drives a multiple a year later, why a ton of money doesn't make you good at valuing catalogs, why song shares don't trade like stocks yet, and what AI licensing will actually pay song owners.Sean Peace: songvest.com · LinkedInChapters0:00 Cold open + intro0:26 Why SongVest is still here5:32 Fewer, bigger buyers9:03 The multiple, a year later13:53 Who pays 25x for Rain On Me16:23 Payouts and getting out21:39 Will it ever trade like equities23:41 AI money and fake streams28:38 QuickfireRecoup designs and builds the systems behind conversations like this one: catalog data, royalty operations, and AI agents for labels, publishers, funds and management companies. recoupable.dev








