How I'm Developing ClueDay as a Product Manager

April 27, 2026 · 4 min read

ClueDay case file artwork representing the build process

Prompted on Lovable for the first draft, GitHub as the bridge, and then Claude Coded my way to the end.

There are a few benefits to being chronically online on LinkedIn and X. I knew all the conversations around AI even before starting. And as a product manager with an engineering mindset I had the confidence to build something. Two reasons why now was a good time to build a word game:

Cheap user feedback: The cost of being wrong is lower than before, and each piece of feedback is easy to act on (add an archive, plug in a dictionary API, show stats). Getting a game in front of people within days would be key.
Taste needs time: You cannot be a novice in a field and expect to have a strong product intuition. It’s built over time. Doing it for the kind of games I have spent years playing came easy.

ClueDay started inside Lovable. I dropped in reference images to set the visual direction, and iterated through prompts until I had the basic game mechanism going.

Prompting screenshot on Lovable — Categorizing the prompts on Lovable.

Connecting to GitHub

The first step was connecting Lovable to a GitHub repository. That one move changed the whole workflow. Now I had a real codebase I could pull locally, version properly, and work on with other tools. I brought it into Claude Code and started from there.

The two-way connection meant changes flowed both ways — I could see live updates on Lovable while working in Claude Code.

Switching to the terminal

I wasn't using the terminal at first, but that changed pretty quickly. Once I was in Claude Code it made more sense to run things directly for checking what was happening under the hood.

Claude Code being used in the terminal to run commands and check the codebase — Claude Code in the terminal, running commands and checking the codebase.

What each tool is good at

Lovable earns its place on UI work. It is fast and the results look great. But for UX changes and anything involving logic or behaviour, I strongly prefer Claude Code. Lovable also burns through credits fast so for the volume of iteration ClueDay needed, Claude Code was the sustainable choice.

Finally, I was able to connect my GitHub repository to Vercel and since I bought the domain too on Vercel, it was an easy connect.

ClueDay — The cases are ready for you to solve them, word detective.