February 2026 · By Jesse Meria
I run a cafe in northern Michigan. Charlevoix. Small town, seasonal economy, four-month peak season that funds the rest of the year. The kind of place where you handle everything yourself because there is nobody else to do it.
One of those things is music. A cafe needs a soundtrack. For years I paid for a commercial music license. It worked fine. The music was fine. But it was always someone else's curation, someone else's taste. I wanted something that actually matched the space I had built.
When I started building this in 2025, I just wanted better music for my cafe. I had found Suno in late 2024. The idea was straightforward: describe the music you want, and AI generates it. I tried it. The first few tracks were rough. But the potential was obvious. If I could get the prompts right, I could create an entire custom soundtrack for the cafe.
So I started generating. A lot. Dozens of tracks a week, then hundreds. I was trying everything. Lo-fi jazz for morning service. Acoustic indie for the afternoon crowd. Warm bossa nova for summer evenings. Some of it sounded great. Most of it sounded generic.
The problem was never Suno itself. The problem was the prompt.
I tried ChatGPT to write my prompts. I tried every free Suno prompt tool I could find. They all had the same issue: they didn't understand how Suno actually works.
They would suggest style tags that Suno ignores. They would write vocal directions that were too vague to be useful. They would exceed character limits. They would combine descriptors that sound good in English but produce garbage in Suno's model. "Ethereal yet powerful ambient jazz fusion" reads well. Suno doesn't know what to do with it.
I was spending more time fixing prompts than creating music. And I was generating enough volume that the inefficiency mattered. Each bad prompt was wasted credits and wasted time.
I'm a builder. When a tool doesn't exist, I build it. That's how I approach everything.
I started documenting what actually worked. Which genre tags Suno responded to. Which vocal descriptors produced consistent results. How structure tags needed to be placed. What character limits meant in practice. I built a system around those findings.
That system became HookGenius.
The core idea is simple: HookGenius doesn't paste your words into a template. It has a full AI songwriter that absorbs your direction and writes from scratch. Over a year of development and thousands of generated tracks later, the system understands how Suno actually works — not how it looks like it works from the outside.
Every generation runs through genre-specific prompting rules built from real Suno output. It handles the structure tags, the vocal direction, the character limits, the energy descriptors. It uses Claude Sonnet for every generation because I refuse to compromise on quality. The difference between a good AI model and a cheap one is the difference between a track you keep and a track you delete.
The tracks I generated started piling up. Hundreds of them. A lot were genuinely good. Good enough that I was playing them in the cafe every day and customers were asking about them.
That led to Puana, my curated AI music library for businesses. The tracks I generate end up in Puana. It's the same pipeline: I use HookGenius to create the prompts, generate in Suno, curate the best results, and make them available to other businesses that need a soundtrack.
HookGenius and Puana are two sides of the same coin. One creates. The other curates. Both exist because I needed music for a cafe and the existing tools weren't good enough.
HookGenius has grown well beyond what I originally needed. It covers 50+ genres with purpose-built prompting rules. It has 180+ free guides covering everything from style tags to pronunciation fixes. It has Sound DNA for channeling artist influence. Album Mode for generating cohesive multi-track projects.
But the core hasn't changed. It's still a tool built by someone who uses Suno every day, for someone who uses Suno every day. No bloat. No features for the sake of features. Just prompts that work.
Because it matters where a tool comes from. Tools built by people who use them are different from tools built by people chasing a market. I built HookGenius because I was tired of generic output. I keep building it because I still use it every day.
If you're spending too much time writing Suno prompts and getting inconsistent results, that's exactly the problem HookGenius solves. Five free generations. No credit card. See if it makes a difference.
5 free generations. Purpose-built for Suno. See the difference.
Generate Your First Song →