Most attention in AI music still goes to the generators themselves. Suno, Udio and similar platforms matter because they can turn an idea into finished audio.
But once a music platform becomes useful enough, another market can start forming around it. Developers begin building tools for the parts of the workflow the main platform does not handle particularly well.
HookGenius, built by Jesse around the Suno workflow, is one early example. It does not generate the final audio. Instead, it focuses on preparing the lyrics, structure, vocal direction and style information that goes into the generator.
The interesting question is therefore bigger than HookGenius itself. Could generative music models become platforms around which a new layer of music software is built?
DAWs already created a market around the core tool
There is a useful precedent in digital music production.
When Steinberg incorporated Virtual Studio Technology into Cubase in 1996, the DAW became more than a closed production environment. VST created a framework in which additional software effects could operate inside the host, while later versions expanded the idea to virtual instruments. Steinberg's company history dates the introduction of VST in Cubase to 1996.
That helped support a much larger plug-in market around the DAW itself. A producer might use one central program while relying on third-party companies for synthesizers, compressors, EQs, reverbs, restoration tools and mastering software.
The DAW did not need to be the best tool for every individual job. Its value partly came from becoming the environment around which specialized tools could develop.
Generative music could move in a similar direction.
Suno can generate the audio, but that is not the whole workflow
Jesse discovered this while trying to build a large music catalog for a café.
He says existing business music services left him choosing between music he did not particularly want to hear and better catalogs that became repetitive. Suno offered another option, so he set himself the goal of creating a few hundred tracks.
That eventually became thousands.
Along the way, Jesse became less interested in simply generating more music and more interested in what had to happen before the generation.
“Suno renders the song. Someone still has to write it.”
Suno already gives users considerable control over their input. Creators can provide lyrics and increasingly detailed instructions around style, structure and delivery.
HookGenius tries to specialize in this pre-production layer. Jesse describes it as preparing lyrics, titles, structural tags, vocal direction and style prompts that can then be used with Suno.
That turns the space between an idea and the generator into a potential software category of its own.
Better models can increase the value of better direction
It might seem that as Suno improves, there should be less need for supporting software.
The opposite can also happen.
If a model becomes better at understanding differences in structure, instrumentation, vocal character and delivery, then describing those things well becomes more valuable. The generator has more capability, but the creator still needs to decide what that capability should be used for.
Jesse describes the goal in terms of intention.
“If we can describe an entire song in words, its structure, its instruments, the vocal persona and how the delivery changes from start to finish, and what you hear is what you typed, that is a very powerful way to produce music. That's intention, not luck.”
This connects with a pattern we have seen in several of our interviews.
Vitalii Klimenko argued in AI Music and the New Burden of Choice that easier generation creates more options but also more decisions. Brian Myk similarly discussed the growing importance of judgment in What Skills Should Artists Value in the Age of AI Music?.
HookGenius approaches the same problem earlier in the process. Instead of helping someone sort through dozens of generations afterwards, it tries to make the creative direction clearer before those generations exist.
“The better you articulate the song, the better it comes out.”
That is Jesse's core bet.
The comparison with plug-ins has an important limit
There is also a major difference between the VST ecosystem and what is currently developing around generative music.
VST is an open plug-in architecture. Third-party developers can build software specifically designed to run inside compatible hosts.
Suno does not currently function like an open plug-in host. Tools such as HookGenius sit outside the platform rather than running natively inside it.
That creates a different kind of market, and a significant business risk.
If an outside company discovers that Suno users badly need better prompting, lyric editing, album consistency or generation management, Suno can potentially build those features itself.
The major platforms are already expanding beyond basic song generation into editing, stems and more traditional production controls. A feature that supports an independent business today could become part of the core platform tomorrow.
For companies building around generative models, finding one missing feature may therefore not be enough. They need to understand a particular workflow deeply enough that creators continue to value the specialized product even as the underlying platform improves.
The opportunity is still much larger than prompting
That risk does not make the surrounding market unimportant. It simply changes what a durable AI music business might need to offer.
Before generation, there can be tools for songwriting, creative direction, artist identity and project planning. Around generation, there may be products for variation, consistency, comparison and organizing large numbers of outputs.
After generation, there are opportunities around editing, mastering, rights information, provenance, catalog management and distribution.
Some of those functions will probably be absorbed by Suno, Udio or whatever platforms eventually become dominant. Others may be too specialized for a general-purpose generator to serve particularly well.
The modern DAW provides a useful lesson here. DAWs became substantially more capable over time, yet specialized plug-ins did not disappear. Independent developers continued to compete by solving particular problems better or serving particular groups of producers more closely.
Generative music could develop a similar division of labor, even if the technical relationship between the main platform and surrounding tools is much less open today.
A software ecosystem is beginning to form
HookGenius is useful as an example because its existence depends on a new assumption.
The audio generator is already good enough.
The business opportunity is therefore not necessarily to build another music model. It can be to make an existing one easier to direct.
Jesse's own experience moved naturally in that direction. What began as an attempt to create a few hundred songs for a café eventually produced thousands of tracks and a separate tool for structuring the work that happens before generation.
It is too early to know which of these supporting categories will become lasting businesses. The major AI music platforms are still adding features quickly, and external developers remain exposed to whatever those platforms decide to build next.
But the appearance of these products is significant in itself.
AI music is starting to create companies whose value does not come from generating the music. Their value comes from making the generator more useful.
The next important AI music product may therefore not compete with Suno at all. It may exist because Suno became important enough to build around.
You can explore HookGenius to see how Jesse is approaching pre-production and creative direction around generative music.
