A good way to start is to look around for things that people or companies are already using (some B2B software tool), and see if there are a ton of people complaining about it (maybe some bad feature, aggressive pricing model, customer support etc.). If so then you can start building an open-source version of it, since you already know what are the key features to build and add.
I was building something on similar lines. It is a semantic search engine for codebases using Openai's embedding and FAISS https://github.com/rahuldan/codesearch
Try taking cold showers, the adrenaline rush helps me get awake and be active. Plus regular exercise also helps towards a general good mood throughout the day.
I am working on an open-source natural language search for codebases. I thought it might be interesting to be able to search for relevant classes, functions etc. when the codebases are very large. Then further I am planning on extending it to searching for git commits, PRs, issues etc.
It has a collection of blogs for building various small projects to learn different languages.