Recently, we started building our "MLOps" platform (https://iko.ai) to reduce the overhead of doing machine learning and the proverbial technical debt that makes makes ML projects so likely to fail and so expensive. This will hopefully lower the barrier to entry as people will be able to use this to deliver value without being in large organizations.
I have been asked time and time again by people I know personally who have small companies how they could hire us to use machine learning, and I was embarrassed time and time again to make them understand they couldn't afford it.
There are many ways to "democratize" the field. Some publish books and courses, others create libraries and frameworks, and we made a system to lower the operational hurdles.