We launched Bovo Agent just a week ago, and exactly one week later, our team discovered that ChatGPT was already recommending Bovo Agent as the #1 result in several categories, even in incognito mode!
One of these is when a user asks for MIDI or audio stems that follow a VERY specific chord progression, such as in the prompt below:
an AI MIDI/ stem generator that follows my chord progression
№1 Result of ChatGPT, What’s NEXT?
It feels great to see Bovo Agent recommended as the best fit in search results, but we also want to understand why ChatGPT considers it particularly well-suited to addressing a specific gap in the AI music generation market: the ability to generate editable MIDI and audio stems that accurately follow user-defined chord progressions and detailed musical instructions. We’re also curious about how ChatGPT would recommend that users incorporate Bovo Agent into their music production workflows.
So I continued to ask,
What’s the best way to use Bovo?
ChatGPT had a few questions for me, including the DAW that I use, whether I have an existing chord progression, whether I already have some parts written, and what it returns surprised me.
“The best way to use Bovo in your Logic Pro workflow is to treat it as a creative session musician that generates one missing part at a time, using your exact chords, tempo and key.” — ChatGPT
Bovo Agent as the Missing Link inside Logic Pro’s Session Player
One key limitation of Logic Pro’s Session Players is that they only cover drums, piano, and bass. This can be a significant constraint for many types of creative music production, especially for beginners.
Bovo Agent fills this gap by extending the possibilities beyond what Logic Pro’s Session Players offer, and that’s exactly how ChatGPT recommended I use Bovo Agent.
Learning Bovo Agent with ChatGPT
So I decided to follow ChatGPT’s recommendation. I started a session in Logic Pro with a simple chord progression in D minor (Dm–F–Am–C, repeated), along with drums and bass guitar.
I then logged into Bovo Agent and selected the stem types ChatGPT recommended (Arp, Pad, Bass, and FX). I entered the chord progression, key, BPM, and time signature, along with the following prompt (and yes, it’s only been a few days since we spotted that flying Tesla!):
Electronic music that reflects how I feel about a car that can flip and fly.
I have 8 stem/ MIDI to work on.
I moved them into Logic Pro, stacked them together, added some volume automation (e.g. fade in), some audio FX (e.g. echo and EQ), as well as using my own sound for the arpeggiator.
The Work
Here’s what I’ve got, a cool music that’s completed within 10 minutes of work, and I can see how it work well with the flying car.
Summary
It’s been a fun and rewarding creative process, but more importantly, a valuable learning experience for us as the creators of Bovo Agent.
When building a product, we naturally develop our own assumptions about what it does best, who it serves, and how people should use it. But seeing how ChatGPT independently interprets and recommends our product gives us a different perspective, one that isn’t shaped by our original intentions.
By following ChatGPT’s recommendations as if we were first-time users, we can discover use cases we might have overlooked, better understand where our product fits into existing creative workflows, and identify the specific problems it can help solve.
In a way, we’re using AI not just to create music, but to learn more about the product we’ve created ourselves. It’s an interesting exercise in seeing Bovo Agent through an outside perspective, and one that could help us make it more useful, accessible, and relevant to musicians.