My mind is officially blown

Hi,

About 2 weeks ago I discovered a relatively new AI platform called String and used it to build another Spotify tool. It took me less time than the last tool I built, was easier, and it built something more robust.

The future is changing in front of our very eyes - I thought it was worth dedicating this week’s newsletter to my experience building it.

Let’s dive in ⬇️

A few months ago I used AI to build my first tool called the Spotify Metadata tool. It provides the metadata and artwork for a specific release (song or album) in a PDF by copying and pasting the Spotify URL to that release.

It was born out of necessity - I found that in my day to day work I was constantly looking for the ISRC codes of specific songs. I was digging into the distribution backend of that particular release and it was always taking me longer than it should have.

I used a combination of ChatGPT and Cursor, toggling back and forth between both platforms to create the tool. ChatGPT would generate the prompt and code for Cursor, then Cursor would actually build the tool.

The tool isn’t overly robust but I was proud that I built something, especially as someone with zero programming experience.

I recently had an idea for a more developed, scaled version of that Metadata tool - instead of providing the metadata for just a specific release, what if there was a tool that could provide metadata for an entire artist’s catalog with the same ease?

I’m thinking ISRC code, UPC code, copyright line…even the artwork for every release in an artist’s catalog.

After watching the Greg Isenberg podcast (a podcast where the host Greg interviews innovators in emerging tech, AI, etc) and seeing an episode about a platform called String, I decided to give it a try.

A few minutes into using it and I was mind blown.

Within one screen, I was able to tell the AI exactly what I wanted to build and watched how on the other side of the screen, the AI verbalized what was happening in real time. Almost as if the left side of the screen was the technical side of the brain (giving me the code and tangible next steps I needed to do) and the right side of the brain was narrating what was happening.

When I built the last metadata tool, I toggled back and forth between ChatGPT and Cursor - I would prompt ChatGPT for code and explain what was happening within Cursor, but the workflow was fractured.

In String, the code was actually built in the same platform I was prompting and all of the manual things that I previously had to test in Cursor, String did for me automatically in platform.

In Cursor, I would have to test specific parts of the tool’s code to ensure they worked properly. String did it for me automatically and when something didn’t work properly, the platform solved the issue by itself without needing to be told.

After what felt like a very short amount of time (no exaggeration, this was less than 2 hours), I built the Spotify Catalog Tool. Here it is: