Personally, I'm doing a web service for image resizing + CDN aimed solely at Polish developers, with the premise that some will prefer having a service in Polish that focuses on Poland (more CDN PoPs in Poland, for instance—I have 2 instead of 0) and would use my service over existing ones. No idea if this is going to work.
I've experimented with the Google Vision API. I built a service to do OCR via an API in a few hours, and it's 100 times better than existing solutions that charge money (like https://www.abbyy.com/en-us/mobile-ocr/). Mainly, it recognizes really small text and handwritten/weird fonts. Of course, it's not me who's good—I'm just using better stuff, but if I made that a SaaS I would have the best OCR API on the market until someone does the same thing.
My girlfriend has a kid, and has to put up with being in a Whatsapp group chat with all the moms, just because every once in a while someone lets the others know that the school is closed for some reason or things like that. However, she constantly gets notifications for small talk, or things like "me too!", while there's nothing important going on. There could be a service that allowed participants to flag messages as "important", "medium", or "unimportant" (or up/downvote them?), and one could decide to receive notifications only for certain types of messages (for instance, "important" only, "medium" an up).
There are so many you can do. Just look around you for problems you can solve.
The best is solving an actual problem you personally have, or someone close to you has. This way you can gauge how important that problem is, and if your solution is good. Hopefully other people will have the same problem and will pay for solving it.