- Jun 24, 2026
- 5
- 9
I had a small internal tool that I'd built a while ago for planning and mapping ideas. It already relied on a few Google services, so I figured I'd see how far I could improve it with the AI tools available today.
I started by asking ChatGPT about the Apps Script side of things, then took that foundation over to Claude to help organize the structure. After that, I ran everything through DeepSeek to modernize parts of the code and clean up a few rough edges.
While testing, I realized the script was still calling an older Gemini model. After digging through Google's AI Studio documentation, I found the updated image generation setup and switched everything over to the newer configuration. It only took a few changes, but it made a noticeable difference.
I also learned a lesson the hard way. I was a little too generous with Google's branding around the project and quickly received a reminder that I probably shouldn't be using their name the way I was. A quick redesign and some rewording solved that problem.
The good news is that the project has already picked up a few sales on Gumroad and some really encouraging five-star reviews. It's nothing life-changing, but it's nice to see something that started as a personal utility becoming useful to other people.
Next on my list is writing proper documentation, improving the interface, adding more model options, and working through some great suggestions I found in the AI community. There's still plenty to build, but it's finally starting to feel like a real product instead of just another side project sitting on my hard drive.
Hopefully I can keep the momentum going. Wish me luck!
I started by asking ChatGPT about the Apps Script side of things, then took that foundation over to Claude to help organize the structure. After that, I ran everything through DeepSeek to modernize parts of the code and clean up a few rough edges.
While testing, I realized the script was still calling an older Gemini model. After digging through Google's AI Studio documentation, I found the updated image generation setup and switched everything over to the newer configuration. It only took a few changes, but it made a noticeable difference.
I also learned a lesson the hard way. I was a little too generous with Google's branding around the project and quickly received a reminder that I probably shouldn't be using their name the way I was. A quick redesign and some rewording solved that problem.
The good news is that the project has already picked up a few sales on Gumroad and some really encouraging five-star reviews. It's nothing life-changing, but it's nice to see something that started as a personal utility becoming useful to other people.
Next on my list is writing proper documentation, improving the interface, adding more model options, and working through some great suggestions I found in the AI community. There's still plenty to build, but it's finally starting to feel like a real product instead of just another side project sitting on my hard drive.
Hopefully I can keep the momentum going. Wish me luck!