Books Have an Invisibility Problem
A conversation with Giacomo D'Angelo, CEO of Amlet
A clarity arrives when two previously separate problems suddenly reveal themselves as one. For Giacomo D'Angelo, CEO of Amlet, that moment came when he recognized a fundamental disconnect in how the publishing and AI industries were approaching the same challenge from opposite directions.
"The turning point was realizing that books do not exist in the world of AI unless they're registered in a central, machine-readable TDM system," Giacomo explains. "AI models do not understand ISBNs, copyright pages, or publisher websites. Without a structured, standardized registry, even the highest quality human-made works remain completely invisible to AI developers."
The Invisibility Problem
Publishers have spent decades building sophisticated metadata systems. ISBNs precisely identify each edition. Copyright pages clearly state ownership and rights. Publisher websites catalog entire backlists with detailed descriptions and purchasing information.
Yet none of this infrastructure speaks the language that AI systems understand.
An AI model training on web-scraped data encounters a book as fragments. It finds a PDF without context, excerpts scattered across websites, summaries and reviews that reference the work but don't identify the rights holder. The careful metadata that publishers rely on simply doesn't translate into the technical protocols that AI companies use to discover and license content.
This creates a paradox: The publishing industry's most valuable works, the ones that are professionally edited, fact-checked, and legally cleared, become harder for AI companies to license properly than amateur content posted directly to social platforms with clear terms of service.
The Regulatory Catalyst
Giacomo noticed that at the same moment publishers were grappling with AI's appetite for training data, regulatory frameworks began demanding exactly the kind of infrastructure that didn't yet exist.
The EU AI Act made it clear that both publishers and AI companies would need shared systems to handle permissions, opt-outs, and licensing in ways that could be audited and verified. Vague website notices and one-off licensing deals wouldn't meet the standard for transparency and traceability that regulators were establishing.
"Amlet was created at this intersection," Giacomo says. "The practical gap, where books are invisible to machines, and the regulatory push demanding clarity and traceability. A global TDM registry is the missing piece that allows human creativity to be recognized, respected, and used responsibly in the AI economy."
Infrastructure, Not Intermediation
What makes a registry different from a traditional licensing marketplace? The answer matters more than it might initially appear.
A marketplace connects buyers and sellers, taking a cut of transactions. A registry provides foundational infrastructure that makes the market possible. It’s like how DNS makes the internet work, or how ISBN systems make book distribution scalable.
Amlet's TDM Registry doesn't sit between publishers and AI companies negotiating deals. It provides the technical layer that allows books to declare their AI usage preferences in a format that machines can discover and respect. The ISCC (International Standard Content Code) technology at its core creates persistent, verifiable identifiers for digital content - solving the invisibility problem at its root.
Publishers register their catalogs once. AI companies query the registry to discover which works they can license and under what terms. The infrastructure handles the technical complexity while preserving each party's ability to negotiate the business terms that make sense for them.
What Changes When Books Become Visible
Once books exist in machine-readable form within a standardized registry, several things become possible that weren't before:
Discovery works both ways. AI companies can find high-quality, legally-clear content to license. Publishers can understand which of their works are being sought for which types of AI applications.
Preferences persist. When a publisher sets TDM attributes for their catalog, those preferences follow the content regardless of where it appears on the internet. The ISCC fingerprint identifies the work even when metadata has been stripped away.
Compliance becomes verifiable. Rather than relying on promises and periodic audits, AI companies can demonstrate that they checked the registry before training. Publishers can prove when they declared their licensing terms. Regulators can verify that the system is working without needing to audit individual deals.
Scale becomes manageable. Instead of negotiating separate agreements for every use case with every publisher, AI companies can discover licensing terms programmatically. Publishers can set permissions once and have them apply across the entire AI ecosystem.
The Moment Becomes a Movement
Giacomo and Titusz Pan, the inventor of the ISCC technology realized the practical invisibility of books to AI systems has grown into something larger: A recognition across the industry that the post-scraping era requires different infrastructure than the open web era did.
Just as the music industry eventually built systems like ASCAP and BMI to handle the complexity of tracking and compensating performance rights, the publishing industry is beginning to understand that AI licensing requires its own foundational infrastructure.
The books were always valuable. The AI companies were always willing to pay for legally-clear training data. What was missing was the translation layer such as a registry that makes published works visible and actionable in the technical protocols that AI systems use.
That's what Amlet provides. Amlet is not a marketplace (although it will offer a licensing platform), rather, it is the infrastructure that makes a well-functioning market possible.
As Giacomo puts it: "A global TDM registry is the missing piece that allows human creativity to be recognized, respected, and used responsibly in the AI economy."
The moment of realization becomes a movement when others recognize the same truth: invisibility isn't a business model. Infrastructure is.
Amlet provides the TDM Registry that makes copyrighted content discoverable and licensable in the AI economy. Learn more about how ISCC technology enables content fingerprinting and rights management at amlet.ai.