AI is reshaping the music industry not just in the studio. It is also reshaping the legal work behind it. Mariia Saveleva knows both sides of this transformation. After working in Business and Legal Affairs at Sony Music, she founded Lexi, an AI-powerd platform that helps music companies, artists and legal professionals navigate complex contracts and rights issues more clearly and efficiently. 

In our conversation, Mariia talks about the legal questions surrounding generative AI and music, the growing role of licensing, and the limits of automation. She also explains why human trust and judgement remain essential in an AI-native legal world – and how her work with Lexi led to the new creation of Aira Law.
 

MTH: Hi, Mariia! First things first: What’s currently on your playlist – and does it have anything to do with your work?

Mariia: Oh, my playlist is a pretty wild mix. If you saw it without knowing me, you’d probably have a few questions. 

This morning it was Stop This Flame by Celeste and I’m Still Standing by Elton John. I’ve just come back from Reeperbahn, so apparently the current theme is: slightly overexcited, slightly exhausted, still standing. 

MTH: The discussion around music and AI is charged. In July, the Munich Regional Court ruled in favour of GEMA in an ongoing legal dispute with AI generator Suno. How did you receive that ruling – as someone who is both a lawyer and an AI founder?  

Mariia: To me, the ruling was not particularly surprising. We are dealing with a completely new type of technology, and the legal framework has not fully caught up yet. There is still a lot of uncertainty around how existing copyright and related-rights rules should apply.

At the same time, technology is already here and already changing the way music is created, how businesses operate and how people work. We cannot turn back the clock or return to the previous rules of the game. 

So, the real question is not whether we accept AI, or how we can prevent AI companies from developing and using these technologies, but how quickly we can build clear and workable rules around them: what can be used, which rights are involved, when a licence is required, who gets paid and on what terms.  

MTH: The question remains how the music industry will handle the issue going forward. For example, Warner Music had also taken legal action against Suno – and ultimately settled with a licensing agreement. Do you think licensing is the direction the industry is heading, or will it take many more court cases to get there? 

Mariia: I think we will see both. Court cases are part of defining the boundaries, especially while the rules are still unclear. At the same time, the industry needs workable commercial models, and that is where licensing comes in. 

For me, it is similar to what happened when new distribution or streaming models emerged. The technology creates a new form of exploitation, and then the industry has to define which rights are involved, who can authorise the use and how the value should be shared. 

I expect litigation and licensing to develop in parallel until there is a much clearer framework.

MTH: With your startup Lexi, you offer a SaaS solution explicitly aimed at legal professionals in the music industry. What is the idea behind the platform and what problem does it solve? 

Mariia: I would actually broaden that a little. Lexi was built not only for lawyers, but for music companies, in-house teams and professionals who regularly deal with contracts and rights, even if they are not lawyers themselves. 

Lexi uses AI to reduce up to 70% of the repetitive manual work around legal processes in music. It can review and compare contracts, identify risks and missing provisions, support redlining, explain complex clauses in plain language, and work across larger sets of documents and rights information. 

The bigger idea is simple: technology should take over as much of the manual work as possible, so people can focus on the decisions that actually require their expertise. 

MTH: AI tools and generative AI are becoming established in music production. The legal questions that inevitably arise from this go beyond authorship alone. What does it take to meet this development head-on? 

Mariia: I think the first step is to stop treating AI in music as one isolated copyright question. It affects the whole ecosystem of rights: copyright and related rights, licensing, personality and voice rights, consent, remuneration, contractual warranties and liability. What we need now is much more clarity around who has the right to authorise a particular use, what exactly is being licensed, how that use is documented and how value is shared. At the same time, companies need practical processes for deciding what can be automated and where human legal judgement is still required. 

Technology is moving faster than regulation, so I do not think the industry can simply wait for legislation to answer every question. We need to start building workable contractual standards, licensing models and safeguards now, and refine them as the legal framework develops. 

MTH: Advances, recoupment, cross-collateralisation: music contracts are often highly complex. What are the most common pitfalls for artists that Lexi uncovers in contracts? 

Mariia: It is difficult to give one universal list because different types of music agreements create very different risks. For me, one of the most valuable things is simply helping people understand what they are signing. Lexi has an “explain” function that translates complicated contractual language into much clearer terms. 

During my legal career, I have seen many artists come to a lawyer only after signing a deal and say, “I didn’t realise this was what the contract meant.” At that point, your options can be very limited. You cannot go back in time and tell them not to sign it. So sometimes the biggest value is not finding one dramatic red flag. It is making sure the person actually understands the economics, rights, obligations and consequences before signing. 

MTH: The music industry is global, and contracts often need to hold up across multiple jurisdictions. How does Lexi handle that – and where does the AI reach its limits? 

Mariia: We essentially work with two layers. The first is the international music-industry layer. A lot of deal structures, terminology and commercial concepts are shared across markets. The second is jurisdiction specific. The legal framework can differ significantly between countries and, in the US, even between individual states. So, both layers are important: the broader industry logic and the specific legal rules that apply to the particular matter. 

AI can bring those layers together and make the first stages of analysis significantly faster. But I strongly believe that, at least at the current stage of AI development, human-in-the-loop is essential. AI is excellent at reducing repetitive manual work, but legal judgment, especially where local law, ambiguity or commercial context is involved, still needs a responsible human. 

Over time, AI can also learn the legal memory of a specific company: how it handles recurring issues, what positions it normally accepts and how previous matters were resolved. That will make it increasingly powerful, but accountability still matters. 

MTH: Alongside Lexi, you are also building another venture: Aira Law. What is the idea behind it – and what sets it apart from Lexi?

Mariia: Lexi and Aira are closely connected, but they solve the problem at two different levels. Lexi is the technology layer. We started by building AI specifically for music legal workflows: reviewing agreements, structuring legal information, working with rights data and reducing repetitive manual work. 

As we spoke with more music companies and tested the product, we realised that many of them were not actually looking for another tool to use themselves. What they wanted was a faster and more efficient way to get the legal work done. 

That is how Aira Law evolved. Aira is an AI-native law firm for music industry. We use technology – including what we developed through Lexi – inside the legal service itself. AI accelerates the repetitive work, while experienced lawyers remain responsible for the legal judgment and final result. 

So, the distinction is simple: Lexi is AI that helps automate legal work, while Aira is the law firm that uses this technology to deliver legal services faster and more cost-efficiently. 

MTH: Speaking of AI-native law firms: do you think AI solutions could, at some point, replace legal counsel entirely – for example for independent artists with smaller budgets?

Mariia: It is a great question, and we discussed it quite a lot at Reeperbahn last week. Right now, I do not think the main limitation is simply technical capability. It is also trust and accountability. If somebody receives legal advice that affects their rights, career or money, they want to know who stands behind that advice. That is one of the reasons we are building Aira the way we are: AI makes the work faster, but a lawyer remains responsible for the result. 

Could that change in the future? Absolutely. AI is developing so quickly that I am very cautious about making predictions five or ten years out. Maybe some of my assumptions today will look completely wrong in just a few months. But right now, I see AI as making lawyers significantly more effective and expanding what they can do. It can transform the speed and efficiency of legal work, while the human remains the source of trust, judgment and accountability. 

MTH: Before founding Lexi, you worked as Head of Business and Legal Affairs at Sony Music – one of the biggest players in the music industry. What did you experience there, and at what point did you decide to start your own company?

Mariia: My experience at Sony was genuinely amazing. I learned an enormous amount there and worked with fantastic people. Sony Music is still very close to my heart. At some point, I simply felt ready to build something of my own and take everything I had learned into a new direction. 

I first started working independently as a music lawyer. Then, through my own legal work, I became increasingly interested in AI and started automating parts of the processes I was doing manually. That gradually became Lexi. So I would not say there was one dramatic moment when I decided to become a startup founder. It evolved quite naturally from practising law, seeing repetitive problems and becoming slightly obsessed with how technology could solve them. 

MTH: Building a startup for the first time is rarely easy – regardless of the industry. What were your biggest personal challenges, and is there anything you would do differently today? 

Mariia: Probably three things: validate earlier, focus harder, and find your people sooner. 

The first one was a mistake I made myself. When you come into a startup as an industry expert, it is very easy to think: “I know this market, I know the problem, so I know what people need.” But that can be completely wrong. 

You need to speak to as many potential customers as possible before you build too much. The problem you think is huge might not be important enough for anyone to pay to solve. Or maybe they do not need the sophisticated system you imagined; maybe they need one simple button that solves one painful problem. I definitely over-engineered things before learning that lesson. 

The second is focus. During my first months, I was involved in everything: events, conferences, networking, programmes. I had a very active social life, met lots of interesting people and was constantly busy. Then I realised that while I was having a lot of fun, the company itself was moving too slowly. 

Since then, I have become much stricter. If something does not clearly help us build, sell or learn something important, I usually skip it. There are plenty of things in the startup world that are great for your ego but not necessarily for your company. 

And the third is the team. I would try to find the right people around me earlier. Building a company can be incredibly difficult and lonely. Finding people who genuinely share responsibility, challenge you and keep moving with you is hard, but once you do, it feels like half the battle is already won.

MTH: Lexi is part of the MediaTech Hub Potsdam ecosystem. In what ways have you benefited from the regional network and the MTH Accelerator?

Mariia: MTH was actually my first accelerator, so it has a special place for me. The mentorship was probably the biggest value. The founder I was when I entered the programme and the founder I am now are almost two different people in terms of experience and understanding. I made mistakes, learned a lot and had people around me who could challenge my assumptions and help me move faster. 

The network has also been extremely valuable: other founders, mentors, events, conferences and introductions that I would probably not have had access to otherwise. And, especially connected to what I said about entrepreneurship sometimes being lonely, having a strong ecosystem around you makes a huge difference. The MTH team has been incredibly supportive throughout that journey. 

MTH: If you had to give one piece of advice to future founders, what would it be? 

Mariia: I know you asked for one, but I have two. 

First: talk to potential customers as much as possible. And remember that someone saying “great idea” is not validation. Real validation is when someone is willing to spend time, change behaviour, test the product or, ideally, pay for it. 

Second: stop comparing your timeline to startup headlines. We constantly read stories about companies raising huge rounds or becoming unicorns in record time, but we hear much less about how difficult, exhausting and emotionally demanding building a company can be. 

MTH: Mariia, thank you so much for the conversation!