June 2026 (Revised June 2026) | Teaching Material
By: Zoë B. Cullen and Nicole Tempest Keller
The note examines how generative AI was reshaping how value was captured in the music industry and across other creative industries. While writers and actors had been able to secure AI protections following the 2023 Hollywood strikes, musicians entered the AI era with weaker institutional safeguards. This divergence stemmed from differences in the “unit of work”: scripts and performances could be clearly attributed, while music production was inherently collaborative, leaving many contributors without ownership rights. As AI platforms like Suno and Udio enabled near-zero-cost music generation and AI-generated content increasingly appeared on streaming platforms without disclosure, musicians were sounding the alarm. At the same time, the use of copyrighted recordings as training data remained difficult to detect, allowing for additional value leakage. In response to these threats, musicians increasingly turned to contractual agreements as the primary tool for asserting control. Their struggle raised broader questions about who owned and profited from creative work when AI codified artistic talent.
May 2026 | Case
By: Zoë B. Cullen, Jordan Chan and Nicole Tempest Keller
In 2026 Agile Agency managed elite OnlyFans creators by handling operational tasks, such as fan messaging and social media posts, so creators could focus on producing content. Initially Agile helped creators scale by training human “chatters” to respond to direct messages from fans, maintaining the illusion of personal interaction while maximizing revenue. As generative AI advanced, Agile introduced tools such as AI chat and AI-generated content, which lowered costs and complexity, while matching or exceeding the quality of human output. Yet these AI tools risked weakening the authenticity fans expected, and potentially even more concerning, they threatened to replace human creators altogether. For CEO Luc Webb, the central question was should Agile continue using AI to augment human creators, or move toward managing fully AI-generated personas that could scale intimacy without humans at all?
February 2026 | Case
By: Rembrand Koning, Hyunjin Kim and Nicole Tempest Keller
Fazeshift, founded in 2024, was an AI-native platform built to automate accounts receivable (AR), a function long dominated by spreadsheets, emails, PDFs, and manual work. Drawing on firsthand experience from a prior startup, the founders developed AI agents that stitched together existing finance systems and handled the unstructured, exception-heavy tasks traditional software could not. After joining Y Combinator’s Summer 2024 batch, Fazeshift gained rapid traction, growing to a ten-person team, more than 30 customers, and 20%–30% monthly ARR growth by 2025. As the company prepared to scale, its founders faced a critical question: how should an AI product that replaces labor and accelerates cash flow be priced—by usage, headcount replaced, subscription, or value delivered? Beneath the decision lay a deeper question facing every AI startup: When AI can rapidly learn and automate complex workflows, how should that value be priced—and who decides what it is worth?
January 2026 | Case
By: Debora Spar, Jeff Huizinga and Demilade Oresanya
This case examines KoBold Metals, a California-based mining exploration startup founded in 2018 with the aim of accelerating the world’s clean energy transition through advanced discovery of critical minerals such as copper, cobalt, nickel, and lithium. Combining artificial intelligence, data science, and geological expertise, KoBold sought to revolutionize mineral exploration—a field long characterized by declining discovery rates and limited innovation. By 2025, KoBold had raised nearly $1 billion across multiple funding rounds, backed by Breakthrough Energy Ventures, Andreessen Horowitz, BHP Ventures, and others, reaching a valuation of $2.96 billion. Its flagship project, the Mingomba copper deposit in Zambia’s Copperbelt Province, was believed to be among the world’s largest and richest undeveloped copper resources, and the team had already used its new tools of discovery and probability to identify 70 additional projects sprawled across four continents. Mining was one of the world’s oldest and most vital industries. It was also deeply and inherently political, characteristics that were emerging as newly relevant as the world’s greatest powers moved simultaneously in the 21st century to develop the massive energy sources that artificial intelligence would demand and to provide this energy through cleaner and more sustainable means. Could KoBold’s novel approach transform the business of mining? Or were the founders’ expectations more likely to be pulled back to earth?
December 2025 (Revised February 2026) | Case
By: Rembrand Koning, Nicole Tempest Keller and Hyunjin Kim
In September 2025, Gamma’s co-founders drove past their company’s new billboard on San Francisco’s I‑80—one of many AI slide‑generator ads competing for attention. Founded in 2020, Gamma offered a prosumer‑focused, AI‑powered platform that enabled users to create highly designed, professional‑quality presentations in seconds, with no design experience required. Gamma had grown to $50 million in ARR with only 30 employees, achieving profitability and lifetime negative net burn by building every part of the company—engineering, design, marketing, customer support—around AI tools. This AI‑native approach enabled Gamma to stay lean, move fast, and serve millions of global users with minimal headcount. But the competitive landscape was shifting quickly. Presentation incumbents like Microsoft and Google were integrating AI into their suites, while startups like Manus and Beautiful.AI pushed forward with aggressive innovation. Even foundation model providers like OpenAI were entering the market directly. As Gamma’s founders looked ahead, they wrestled with a critical question: would their AI‑native, product‑obsessed culture be enough to gain a sustainable edge in an increasingly crowded AI market?
