The modern smart phone era, as kicked off by the launch of the first iPhone 10 years ago, is now mature.
At the launch of Google's Pixel 2 phone in San Francisco today, CEO Sundar Pichai acknowledged smartphone features were "levelling off" and said it was hard to develop exciting new products based on hardware alone.
Google said it is in a transition from a "mobile-first" company to an "AI-first" business.
The machine-learning part of artificial intelligence is one of Google's strengths.
Neural machine translation now understands 96 languages and does 2 billion translations per day.
A demonstration that best illustrated the power of the new integrated Google was a live translation featuring a woman speaking Swedish — while wearing wireless Google Pixel Buds headphones — to a person holding a Google Pixel phone and speaking English.
Machine learning (ML) has already revolutionised the photography workflow by taking pictures, adding effects, and most importantly finding images in a large collection.
ML powers the Google Assistant that lets us do Google searches by simply asking questions. The ability to access other big data sets, including maps, means natural but sophisticated questions such as, "Is the chemist on Station Street open?" can be answered faster using speech than a web search.
Google continues to advance software technology to build better machine learning models. Their "AutoML" takes some of the previously required human expertise out of the model building and produces results that improve month by month.
Many questions to the Google search engine are not easy to express in words and work is underway on "Google Lens" which can look at, for example, an advertising flyer and read the phone number, email address or web URL from it and take the appropriate action for the user.
The lenses in this year's update of the Daydream VR viewer have been improved by the application of in-house lens simulation software that allowed the team to vastly increase the number of lens design variables that could be tried compared to previous techniques.
A compact, standalone camera, Google Clips, has built-in machine learning that watches the scene playing out and selects the most photogenic stills or short clips for your library without human intervention.
This ML model is running on the tiny device and demonstrates Google is responding to concerns about privacy from key competitor Apple.
Google Home mini, a compact version of last year's Google Home, will sell for $79.
Like its more expensive predecessor, this one brings the ability to simply talk to the Google Assistant by prefixing your question or command with "hey Google".
About the size of your hand and with a clean, cloth-topped design, these gadgets bring other household devices into the age of speech control and are hoped to replace kitchen timers, weather stations, recipe books, radio, light switches, air-conditioning control and even your bedside alarm clock.
The smartphone era has revolutionised our world. Smartphones are our most used devices but Google and Apple both know the race is on to figure out what will come next.
Apple has shown control of both hardware and software is key to a coherent experience and Google has clearly taken that to heart with the Made for Google products we now see.
Google is strong where Apple is weak — in their robust cloud services and machine learning expertise — and it's clear they are focused on exploiting advantage to chip away at the powerful Apple ecosystem.
Peter Marks is a software developer and is in San Francisco as a guest of Google.