Before the First Note — Part One

What Spotting Sessions Reveal About the Future of AI Music Tools

Part One of a two-part look at the work that happens before a score is written.

After more than twenty-five years composing for film and television, I’ve come to believe that the hardest part of scoring a scene usually isn’t writing the music. It’s getting everyone involved to agree on what the music needs to do.

I’ve been part of hundreds of spotting sessions. Some lasted fifteen minutes. Others took hours. The goal was essentially the same: figure out what each scene needed from the score.

Those conversations were rarely just about music. They were about story, pacing, subtext, character, performance, editorial rhythm, and the emotional experience of the audience. Directors, producers, editors, composers, music supervisors, and music editors were all looking at the same scene through the lens of their own work, but not necessarily seeing or hearing the same things.

Most of the time, they were productive. Ideas built on one another. Assumptions were challenged. The group gradually arrived at something better than any one individual had brought into the room.

But not every idea entered the conversation equally. Some people could describe exactly what they meant. Others knew what they wanted but struggled to put it into words. Some ideas received more attention because they were expressed more confidently. Others never quite found the right language.

Decades ago, we were scoring a national Pasta Roni campaign. The agency and producer were happy with the cue, but the producer had one additional note. He wanted a “crisp entrance of the bassoon” when the product’s hero shot appeared.

There was no bassoon in the cue.

After a few minutes, we figured out which instrument he actually meant, made the adjustment, and everyone was happy. We jokingly named the cue “Bassoonaroni,” although a bassoon never existed in it. The producer’s note wasn’t wrong—he knew what he wanted to hear, he just didn’t have the vocabulary to identify it, and once we understood the intent behind the note, the solution was simple.

Versions of that exchange happen every day in post-production. Creative people are trying to express a shared intention through different vocabularies. Over time, I came to see those conversations as part of the composition itself. The score really began there.

I didn’t initially think about AI as a way to address that challenge; that realization came later. My interest in AI music began where it probably does for most composers—with songs. I wrote a bunch, tried different genres, and explored what the platforms could do. Eventually, I started uploading my own compositions to see where the technology helped, where it failed, and whether any of it belonged in a professional workflow.

One experiment changed the direction of my thinking. I had written the main song for an animated feature musical and produced a rough demo. I uploaded it into Suno and generated thirty different production approaches based on directions I wanted to explore. Most were interesting. Several were very good. Two were exceptional. They didn’t take the song in new directions. They realized ideas I was already exploring, but more fully than I expected. I combined the strongest elements and ended up with something much closer to a Broadway or Disney cast recording than the original demo, in a fraction of the time those same explorations would have taken traditionally.

The bigger surprise was how hearing the music changed the way people engaged with the idea. The writer and director could respond to the result instead of my description or approximation of it in the demo. They could point to a texture, an arrangement choice, a vocal approach, or a moment that worked. Ideas developed more quickly because everyone was reacting to the same thing instead of imagining it differently. That changed the conversation.

From there, I began experimenting with music for picture. I generated cues against scenes and listened less for whether the music was impressive than for whether it worked. Did it support the story? Did it leave room for dialogue? Did it understand the pacing? Did it arrive at the right emotional moment?

I tested those results against footage from shows I had scored in the past. Current platforms still do not reliably recognize timings or hit points in a prompt, so the generated music had to be edited manually to picture. That is a significant limitation. But the process still made it possible to hear and compare musical directions much earlier than a traditional workflow would allow.

It also suggested what this kind of immediate exploration could make possible in a collaborative workflow. Temp ideas could be heard while a scene was still being discussed. Editors could compare alternatives instead of trying to imagine them. Producers could respond while they were still in the room. People could build on one another’s realized ideas rather than descriptions of ideas, allowing iteration that once took days to happen in minutes.

That does not mean the first result would be best. Often it wouldn’t. The advantage is the ability to compare, reject, redirect, and refine. It also makes the creative conversation more immediate, grounding it in something everyone can hear and respond to.

The first creative product in the scoring process is not the cue. It is the shared understanding that comes before it.

Everything else depends on that: the brief, the first pass, the revisions, and the final score. When the intent is clear, the composer can spend more time honing the craft. When it isn’t, everyone can lose days responding to music built on a misunderstanding.

That observation led me to develop a tool called CuePilot™. I’ve built it and have been testing it around this process.

CuePilot starts with the scene rather than the music. It provides a structured way to gather and organize the intent behind a scene, then carries that thinking into both a professional music brief and a prompt for an AI music platform.

It is designed to be used by directors, producers, editors, and other creative stakeholders without requiring musical training or technical vocabulary. They can describe the scene through story, character, emotion, and timing, and CuePilot translates that intent into actionable musical direction.

I’m now developing CuePilot toward a collaborative portal in which each stakeholder can contribute directly. Their contributions would remain identified throughout, creating a record of ideas, differences, and alternatives worth exploring. The goal is to gather the creative intent behind a scene not just from one person, but from everyone involved, giving the team a shared picture of what it is asking for before anyone begins producing answers.

These questions are familiar to anyone who has spent time in a spotting session. What is happening in the story? Whose point of view matters? What should the audience feel, and when should they feel it? Where does the music enter and leave? Which moments need support? What should remain unspoken? What musical directions are worth exploring? What practical limits need to be considered?

Different members of the team may answer those questions differently. That is not a flaw in the process. It is the process. The value is in surfacing those differences early enough for the team to discuss them, build on them, and arrive at one shared understanding of what a scene needs.

CuePilot does not decide which idea is correct. It provides a framework through which different perspectives can become a unified creative direction. It makes priorities visible and carries the resulting intent into the next stage of the work. The purpose is not simply to create a better prompt. It is to help the team work through those differences before a creative direction is put into practice.

Current AI music platforms are not yet built around the realities of professional post-production. They can generate compelling music, but they do not yet function reliably inside a process where cuts change, timings move, themes recur, revisions need to preserve what is already working, and an entire score must maintain a coherent identity.

The next generation of tools will need to honor precise timings and hit points, adapt to evolving edits, preserve continuity across an episode, and maintain a consistent musical identity throughout a project. That is the horizon line. That is the work.

The larger opportunity is to support the full process: spotting, exploration, continuity, revision, collaboration, and delivery.

The strongest scores I have worked on did not arrive fully formed. They developed through conversation, experimentation, disagreement, revision, and eventually alignment. Any tool that improves that process without flattening it could become valuable.

The future of AI in post-production is not just about generating music faster. It is about helping creative teams understand what they are trying to make, giving them better ways to explore and articulate it, and improving the conversations that get them there.

© 2026 Rick Butler. All rights reserved.

Continue to Part Two: Before Anyone Else Hears It