My daughter is three years previous. She broke her toy guitar and needs a brand new one. As an alternative of one other toy guitar, I purchased her an actual Ukulele. It is small, light-weight, and excellent for a small lady. In addition to, it sounds approach higher than the annoying toy guitar.
She liked it. However I am the one who performed with it probably the most. I do not know easy methods to play the ukulele. I used to be studying, and it was enjoyable.
One Saturday morning, precisely at 7:15 I acquired an concept for an app. Have you ever seen karaoke apps? You sing karaoke whereas the lyrics on the display scroll down. I needed to construct one like that for Ukulele gamers (for myself). They’ll see the lyrics and the Ukulele chords scrolling on the display. They’ll sing alongside whereas taking part in the ukulele.
The most effective half: yow will discover Ukulele Chords for nearly any tune ever sung on the planet. It would not should be “Nation Roads” or “Someplace Over the Rainbow” on a regular basis. You possibly can kind in any unpopular tune; the app can nonetheless generate chords for it. I wish to construct it with an AI agent that finds the tune on the web (YouTube primarily) and makes use of that to deduce chords. Thus, so long as the tune you typed within the search bar is someplace on the web, you may discover the chords for it.
I sat down in entrance of my pc, opened Google AI Studio, and advised it what I would like. My spouse needed me to hitch the household for breakfast, however I stored saying, ‘One second, please’, borrowing extra time. By 8:30, I had not solely developed the app, however I additionally deployed it, purchased a site, linked it to my app, and shared the hyperlink on a Reddit thread.
This is what it seemed like.
Ukuflow is a Vibe coded AI powered App for ukulele learners. That is it is Dwelling web page
Ukuflow is a Vibe coded AI powered App for ukulele learners. That is it is Music web page
I’m an expert programmer. I have been coding for the final fifteen years of my life. Belief me, if I needed to develop this all on my own, I would have taken a very good six months. However I pulled it in about an hour, even with out writing a single line of code.
Truthfully, AI is superb.
Later that day, I checked my Reddit put up. I acquired a incredible response. One redditor even stated, ‘Sounds just like the uke app I’ve at all times needed!‘ I slept properly that day, feeling that I lastly constructed one thing the world needed.
The following morning, I checked the stats on Google Analytics. I’ve acquired round 200 individuals who tried my app. That is nice! I then checked the billing tab in my Google Cloud Console. That is the place I felt that what I constructed was a flop.
I’ve spent $52 operating the app only for a day. The AI agent is liable for most of it. Each time a person visits the positioning, they may use this flashy characteristic to seek for songs that they thought they might by no means discover. And that shot the fee sky-high.
My vibe coded AI powered app was stay just for in the future and it spent $52.
You might argue that is regular for an app like this. I agree. However there was a couple of purpose to kill the app.
A enterprise that fails at a unit degree would not earn cash
Earlier than this, I’ve run two blogs myself. I’ve monetized them with show adverts and affiliate hyperlinks. From that have, I can say, typically, you possibly can earn about $10-$20 from each 1000 visits to your website. When you’ve acquired one thing to promote, you might goal for $30-$50. Meaning my 200 guests would have earned at most $4. However at what value? $52!
Unit economics fail, and that does not sound good to me. Even when I purchase an organization’s inventory, I take a look at how a lot it makes from each sale. That is one thing I discovered from studying the works of the best traders the world has ever seen, like Peter Lynch and Warren Buffett.
Some have stated that the price of constructing an app like this for six months would have been recouped. However thoughts that this $52 is the operating value of the app. Not the fee to develop it. I developed it free of charge. The Gemini free credit have been adequate for all the things I needed to do on this app. The operating value, nevertheless, would hold accumulating.
Dropping $48 ($52-$4) a day would find yourself dropping $17k a 12 months until I discover different upsell alternatives.
Is the ache actual?
What am I fixing for?
The flexibility to generate Ukulele chords for actually any tune on the web.
Actually, lots of people would need this app. I figured it out from my very own ache. I liked taking part in the ukulele, and as a newbie, I would like to have one thing like this. AS A BEGINNER!
However what’s going to occur as soon as I grasp it? I in all probability infer the chords myself from my expertise. I do not want an app to do this. Meaning the market I initially envisioned was even smaller. It is not any ukulele participant; it is novices who battle to deduce chords and coordinate singing and taking part in collectively. Execs do not want this.
Okay, however what about these novices? Will they not use it sufficient? They’d. However novices normally are inclined to play a dozen songs which can be universally thought of the ‘beginner-friendly ukulele songs.’ If it is for my daughter, it would be one thing like “Baa Baa Black Sheep” or “The Wheels on the Bus Go Spherical and Spherical.” For grown-ups, maybe the identical previous “Nation Roads.” Widespread songs are adequate for novices. Execs know easy methods to infer chords for different songs.
However after I used AI to deduce chords, I burned credit. Which is why the $52. It is a expensive whistle I held on a genuinely useful Christmas tree.
The issue is actual, however the resolution is overkill.
A bunch of statically generated pages would do the trick as a substitute of a complicated AI agent. And that is precisely what I’ve carried out now. When you go to ukuflow.com at this time, it is a bunch of songs you possibly can choose and play alongside to. Made for novices, by a newbie.
Begin small and iterate.
As somebody who labored on many SCRUM initiatives, I ought to have recognized this higher.
However I additionally suppose that is the smallest experiment I’ve run: failed with solely a $52 invoice, not $5k or $50k. I’ve additionally carried out the course correction of constructing the songs static as a substitute of stay AI technology. However I nonetheless wish to discuss it as a result of now that AI lets anybody construct virtually something, extra might go down a rabbit gap.
The most effective place to begin is testing your vital assumption about your resolution. For my Ukulele app, this could be that folks would wish to see lyrics and chords scrollling like they do in Karaoke apps. Even a single tune web page would have examined this assumption. That is my minimal viable supply. Folks would nonetheless discover it helpful and provides their suggestions. I can resolve whether or not to scale it up with extra songs, add extra options like AI technology, or drop the concept altogether.
Ultimate ideas
At one level in my group, I used to be the one one who coded in Python. Right now, I additionally picked up Java, JavaScript, TypeScript, and some others. However you recognize what? So does my mother (who has by no means written a single line of code).
When AI coding got here in, I immediately misplaced my aggressive benefit. Fortunately, I did not rely solely on coding. That is my first lesson.
Concepts have been at all times cheaper. Execution wasn’t. With AI, though it appears execution has grow to be simpler, planning nonetheless hasn’t. To succeed as a vibe-coder, or at the least to not fail dramatically, we needs to be extra agile. We should always take heed to the top customers much more than we used to, then iterate and take a look at quick. That is the larger lesson strengthened after seeing my vibe-coded app fail.
