AI The buzzword that makes every creative’s heart race. And fair enough. We’re being flooded with “magic” tools that promise a masterpiece at the push of a button. The reality, of course, is more complicated. In this piece, Maik Cox, ECD at Ambassadors, explains how they’ve moved past the hype and are using AI as a versatile co-pilot in music production.
More than a lucky prompt
Wouldn’t it be nice if you could drop an idea into AI and instantly get a track that does exactly what you want? Nice fantasy, but not how it works. If you want to create something that really works, you still need a human who can adapt, understand the composition, and adjust it, especially since you’re often working with visuals that aren’t finalized and may still change. Even though it’s improving, AI music often feels unfinished and underproduced, particularly with acoustic instruments.
Yes, every once in a while you get a prompt that generates a track that’s decent enough, and you can hear the potential in the idea. But reliable, production-ready results? Not so much. And even if the AI spits out something promising, that’s only the start: maybe you need the track to stretch out in the middle, switch tempo halfway through, or end on a different emotional note. If all you’ve got is a single wav file, you’re stuck. Some tools now let you export stems (separate layers), which helps, but compared to a traditional composition you’re still dealing with way less control and lower quality.
And then there’s the legal side. The tools producing the “best” results today are typically trained on huge amounts of copyrighted music, without permission. In the U.S., this has sparked major lawsuits, leaving the legal future uncertain and requiring anyone using AI-generated music to be cautious.
A new teammate
Despite all that, AI has become a fixed part of our workflow. Not as a composer, but as a creative partner. It helps us speed up demos, spark new ideas, and sometimes take us in directions we wouldn’t have found on our own. When used well, it’s a beautiful extension of creativity.
Here’s how we’re putting it to work:
As a source of inspiration: We use models that generate rough sketches. They give us fresh angles or inspiration for alternative routes.
As a smart assistant: We’re testing a new beta program that analyzes our compositions and suggests matching instruments or custom loops. Sometimes it’s valuable, sometimes completely off, but it’s always interesting.
As a vocal shapeshifter: With tools like Vocoda.ai (speech-to-speech), we can turn a single vocal recording into multiple versions with different timbres almost instantly. This lets us quickly test which voice colors work best, and by starting with just one strong input voice, we can slightly lower demo costs.
As an audio post tool: We use AI to filter out reverb, noise, and other imperfections from set recordings with almost one click. It’s not always perfect, but when it works, it saves us a lot of time.
Experience makes the difference
For many creators, AI feels like an intruder in a craft built on personal intuition and skill. And yes, it can feel uncomfortable. At first, it almost felt like we were cheating ourselves: decades of training, practice, and flight hours suddenly condensed, as if by a prompt. But resisting it isn’t the answer, falling behind is the greater danger.
What we quickly realized is that it’s precisely those years of experience that allow you to get something truly useful out of AI. Without that, it’s usually nothing more than a draft that requires (a lot of) work. It’s natural for creators to be cautious at first, but that resistance fades once you start seeing AI as an extension of creativity rather than a threat.
We also stick to a simple rule: use AI in a fair and legal way, so we don’t compromise our principles.
Music x AI breakfast session
Over coffee and croissants, we’ll show you how we use AI in music production. No boring slides or distant predictions, just live music making, hands-on experimentation, techniques that really work, and seeing what happens when things don’t go exactly as planned. You’ll see how we sketch out ideas faster, switch a vocal from male to female in seconds, and let AI suggest new directions.
Sign up here.
