Open-source tool - to develop in the open or not?

I'm (very slowly) writing this open-source tool for machine learning experiment management. I know there's lot of things in the area at the moment but I feel it's a big issue for ML devs/researchers and there isn't a good solution out there yet. I feel like they all lack some level of simplicity that I think I might be able to provide.

This "tool" I'm making has many parts and I want them to work independently (like python logger for results, command line tool for reading results, etc.).

At what stage should I actually open-source it? I feel like I want people to know about the idea and get feedback on it if possible, but I don't want it to look like a joke project since it's in the very early stage right now. I mean, when I browse projects and I see something too early stage I don't even bother following progress...

Where do you draw the line? When to make it public? Should it have one useful part with others to follow or is it good to develop in the open from the beginning?

Of course the plan for later is to offer a hosted version if this proves to actually be useful. Effort won't be wasted on building it since I'm using it.