Back in 2019, I wrote about the need for data hygiene in the run-up to the launch of the Mechanical Licensing Collective (MLC). My argument then was straightforward: If music publishers wanted to operate effectively in a more data-driven environment, they had to get serious about the quality, consistency and accessibility of their information. Seven years later, publishers are asking most prominently: How do we take advantage of AI?
For many music publishers, AI can be genuinely useful. But too many organizations are starting in the wrong place. They begin by asking which vendor to use, which model to test or how quickly they can deploy a new tool. Those are not the right questions. The real jumping-off point is much simpler: What exactly are we trying to fix?
The Real Problem
For most music publishers, the real obstacle is not technology-related. Rather, it is fragmentation — the same issue they were having back in 2019. Throwing more tech at any problem rarely resolves it. In fact, when systems, workflows and responsibilities are misaligned, AI tools do not create clarity. Instead, they amplify inconsistency, highlight contradictions and make weak foundations more visible.
In my experience, three structural bottlenecks commonly appear.
Identity: Which work are we talking about? That sounds elementary, but it rarely is. The same composition, recording or writer may exist under multiple identifiers across different systems. Titles may vary, names may be formatted differently and metadata may be incomplete in one place and duplicated in another. When this is the case, even straightforward matching becomes more difficult than it should be.
Logic: Royalty calculations are rarely just formulas. They are an accumulation of contractual terms, policy decisions, historical exceptions, operational workarounds and institutional memory. Some of that logic lives in software, some of it lives in documentation, and some of it lives only in the heads of the people who have been holding the system together for years. If publishers cannot clearly explain how a rule is applied, they should not expect an AI layer to apply that rule reliably.
Lineage: In publishing, trust depends on being able to answer a basic question: How did we get to this number? That question becomes critical when a royalty payment looks wrong, when two systems produce conflicting outputs, or when leadership wants confidence that a recommendation can be verified. If the path from source data to reported outcome cannot be traced, AI will only augment the problem. For music rights infrastructure to become automated, it must first be explainable.
Trust in and the effectiveness of AI will not come from how impressive a demo looks. It will come from whether someone inside the business can follow the path from data to outcome and understand what happened at each step. AI systems will only be as intelligent as the structures they operate on.
Are You Ready?
To help identify and fix these bottlenecks, I often recommend that publishers begin with an external AI readiness assessment instead of jumping straight into implementation. An outside review creates useful distance from the assumptions that develop over time inside any organization, and it helps surface where systems contradict each other, where responsibilities are unclear and where the business is carrying hidden operational risk. It makes the implicit explicit.
A good assessment is not about slowing innovation down. It is about making better decisions sooner. It should clarify how work actually moves through the organization, where the critical data lives, which dependencies are fragile and which use cases are worth pursuing first. That process should always include stakeholder interviews, a review of data quality and system fragmentation, and a practical analysis of where AI can help versus where it is likely to create more complexity than value. The goal is not a theoretical AI strategy deck, but rather a grounded 12- to 18-month roadmap that leadership can act on immediately. That may sound less exciting than buying a new tool, but avoiding the wrong AI investment is often more valuable than finding the right one a few weeks earlier.
The music industry does not need more AI theater. It needs more operational clarity. The publishers that benefit most from AI will not necessarily be the ones that moved first. They will be the ones that took the time to understand their own systems, align their workflows and build an environment where intelligence can be trusted. AI can absolutely create value in publishing, but only after publishers know what they need the technology to do.
Guy Barash is the founder and CEO of Dotted Eighth LLC, a boutique technology advisory firm working at the intersection of music, data, and emerging technology. In this role, he advises music organizations, tech companies, startups, and others serving the music industry on the data and technology that powers music rights. With a background that includes extensive publishing experience and more than 20 years as a composer, Barash excels at navigating the intricate creative, legal, and technical domains that the music industry contends with day-in and day-out.
