Fixing Publishing for the AI Era: Why We Invested in Cashmere

The publishing industry is undergoing a massive shift as generative AI transforms information consumption, with users increasingly seeking direct answers to their questions rather than links and full publications. This shift has been great for end users but devastating for publishers. Traffic is collapsing, legacy licensing models don’t map to AI-native workflows, and much of the world’s most valuable written knowledge is being absorbed into large language models without clear control, attribution, or compensation — eroding the incentives to create future knowledge.

In our view, the solution is not the current status quo of lawsuits and bespoke licensing deals, but novel technology infrastructure that lets publishers decide how AI uses their content — and get paid when it does. 

Cashmere aims to create a comprehensive infrastructure layer that addresses the challenges of content control and monetization in the AI era. Premium publishers today cannot safely expose their content to AI systems while maintaining granular control over access, usage, and compensation. Once content is indiscriminately used for training, it is effectively irretrievable. At the same time, AI companies do not want to manage hundreds of bilateral licensing relationships or maintain custom integrations with individual publishers.

Cashmere sits precisely at that bottleneck. It is building a neutral, API-first control room that allows publishers to ingest content once into an MCP server (where it is transformed into multiple queryable representations), define exactly how it can be accessed and monetized, and then make it available at inference time to AI products in a rights-cleared and metered way. AI companies get a single integration point for authoritative content; publishers retain control and visibility. 

We believe the opportunity is huge. In an AI-driven world, the most valuable queries are not casual, generic searches. There are many moments when correctness, citation, and trust are mission-critical, in education, research, medicine, law, finance, and professional decision-making. Those answers cannot come from scraped web content or stochastic parroting; they require vetted, premium sources. As AI adoption accelerates across enterprises and institutions, demand for rights-cleared, authoritative content only increases.

An important aspect of our investment thesis is that publishers are hesitant to depend directly on the AI labs due to historical bad behavior and a lack of structural accountability. The trust has already been eroded. Even when labs promise better behavior, publishers have little incentive to hand their IP directly to companies whose core business is building models, not managing rights. Cashmere’s neutrality is the key. By positioning itself as independent infrastructure rather than a buyer, reseller, or model provider, Cashmere can serve as a trusted intermediary for both data and monetization. Meanwhile, on the other side of the market, AI companies get better-quality answers, higher user trust and retention, and a scalable way to integrate premium content without building a publisher-facing organization from scratch.

The Cashmere founding team — Jonathan Munk, Jonathan Woahn, and Wyatt Grantham — is unusually well-suited to solve the premium content problem for AI. They combine empathy for publishers and a deep understanding of the complexities of premium content from their time building their prior venture BookClub. They possess strong technical instincts and a clear-eyed view of how AI is reshaping incentives and economics across the publishing ecosystem. They’re also just kind, earnest, high-integrity people, exactly the kind of founders we love partnering with at Reach.

As Cashmere succeeds, it will become foundational infrastructure — the Stripe or Plaid for rights-cleared knowledge in AI. They are already working with marquee publishers and AI builders alike, including Wiley, Harvard Business Publishing, and Perplexity. We’re excited to lead Cashmere’s $5M seed round, with participation from industry leaders such as Pearson, Ingram, and Naver, and partner with the team as they build the plumbing for authoritative knowledge in the age of AI.