I have a hunch why thats the case. The reason is that past success in transactional business has biased tech industry towards solutions that are transactional. Education is anything but transactional, unless you count selling books as education.
The success of past has put a filter which prevents good eductech companies.
Education is a workflow that spans easily months or years to be effective. A good startup in this area would avoid being transactional, instead it would learn from success of another tech process that is long drawn. That is making good software. The kinds of process that helps to make software development more effective are the kinds of process that would make long drawn learning process effective.
This begins with github for learning lessons which allow people to collaboratively create learning lessons. The best ones created by community would float to top based on approaches similar to guthub stars or npm downloads.
So much about learning effectively is about creating and and managing content flow and collaboration.
Today creating and sharing text, photos, videos, audio is extremely simple thanks to advances in tech and mobile.
Today collaboration in real time over internet is increasingly easier to add to any app. And this is not restricted to chat.
Today algorithms can do great job of helping you discover learning content as thousands and millions are created by people.
All this and more would in my view would create an education platform for tomorrow.
Let me know what you think ?