Charlie Warzel at BuzzFeed has a good analysis of Google's I/O, the new Now on Tap feature in Android M in particular:
If you’re looking to understand what Google hopes to accomplish with its wildly popular Android operating system in the coming years, Now on Tap is a good place to start. It’s a sleek, impressive, and ambitious demonstration of machine learning that sits always at attention, ready to sift through the pile of data your smartphone is constantly producing in order to make your life easier. Now on Tap is the connective tissue between you and your apps and the internet at large — your very own search butler. [...]
Even third-party apps, arguably the only escape hatch out of Google’s walled garden of software and hardware, operate within and must defer to Android’s connective tissue features if they want to take advantage of things like On Tap. Google I/O is billed as a conference for developers, and today Google courted them and made them hoot and holler and clap at a number of features that promise to make their programs better, faster, and more efficient. But almost every single feature on display today was about Google’s play for more control.
But while I think Charlie is right in arguing that Google's initiatives point to the company wanting to acquire more control, of data and of the Android experience, I disagree with his assertion that all of Google's products and projects come together in order to "just work". With the fragmented state of Android — last year's Lollipop is running on less than 10% of active devices according to Google's own data (not including pre Android 2.2, AOSP and forked Android devices) — and the number of OEMs that Google works with, all of what Google announced in the past two days coming together in any meaningful way seems a way off to me.
Parts of the experience work, and are sticky, like the new Photos app. Google Now on Tap also looks like a tremendously useful tool that many Android users (at least of those that get Android M in a timely fashion) will begin to rely on. But Brillo, Weave, Nest, Cardboard, Wear, Android One, Jump and all the other pseudo-random projects Google has do not, in my opinion, form what Charlie dubs "a comprehensive suite of technology". More importantly, we don't yet know which parts work or indeed which parts will even stick around until next year's I/O. Google is making some interesting noises, but it will be a while before we see any of I/O's announcements come to fruition.