Major themes and products announcement @ IO 2019
Google IO is one of the most awaited developer conferences of the year. It’s more amazing to see how a search giant has transformed themselves into a company providing tools and software for developers from mobile to web to cloud.
Amid much fanfare, this year’s Google IO kick-started with their vision and transformation statement -
A company that helps you to find things to a company that can help you get things done
More than just highlighting the key announcements, I wanted to identify the major themes around Google IO and classify products under those buckets.
Below are the major 7 themes that I would like to fit in this year’s IO.
1) On Cloud AI -
1.a) Duplex on the web -
More focus on getting a reservation done by filling the forms automatically by understanding the context from Gmail, calendar, etc through Google assistant. This doesn’t need any custom implementation of the websites.
1.b) Project Euphonia -
More capabilities to understand voice. If someone has a slurred or hard to understand speech they can submit the samples to Google and they claim they can improve voice capabilities.
1.c) Google AI -
BERT — Bidirectional context understanding of the words in a sentence.
Early detection of lung cancer models and flood forecasting models were demonstrated on how Google AI is being used.
Cloud AutoML helps build your own models from your own dataset.
1.d) Firebase -
AutoML vision edge to simplify the classification of images through firebase. Performance monitoring gets web addition apart from existing Android and iOS.
2) Augmented Reality realization -
2.a) AR and Camera -
The camera is more powerful than ever. We could see many of Google’s products use the camera to enhance the experience in a more helpful way.
2.b) Search -
Deep integration of AR in Google products like search. Viewing the objects in 3D directly from search and viewing the objects in your space in AR.
2.c) Google lens -
Upgraded Google lens app that can highlight the popular dish on a restaurant from the menu directly and even pay the bills by splitting the bill with your friend and adding the tip. Just everything happens from your camera.
Translate, listen and search the texts from any image directly from Google lens. It mostly resembles an existing Samsung Bixby feature. However, the most impressing things were all of these features comes under 100 Kilobytes.
3) On-Device Learning (Federated learning) -
Google’s AI products process the data on the mobile device and update the global model to the cloud and the global model will be shared with other devices. This will be a routine process. In this model, we don’t send personalized data to the cloud but just the models as it gets upgraded.
100 GB of voice models reduced to 0.5 GB through breakthrough technology in deep learning. This improves latency by 10 times faster since all of the voice can be processed on the device without hitting the cloud.
On-Device translation API support for 59 languages.
3.a) Personal preferences in assistant -
You can share your preferences of people, places, etc in the updated ‘You’ tab in assistant settings to get more personalized results and also make assistant understand the context clearly.
3.b) Live caption -
On device voice translation without sending it to the server. Reduces lots of latency.
3.c) Live relay -
Phone calls without actually speaking on the phone. Looks like an upgrade to existing Google screen call assistant in calls.
3.d) TensorFlow Lite -
Helps on-machine learning where there is less network connectivity, privacy concerns and reduces latency.
4) Google assistant -
4.a) Content creators -
HowTo Markup and HowTo Template enable content creators to make minor changes in websites and that can give seamless experience in assistant devices.
4.b) App developers -
Full-screen display mode for smart displays using interactive experiences.
4.c) Assistance in the car -
A new driving mode for the assistant. There is no new app needed. It comes with Google assistant coming this summer.
5) New devices launched -
5.a) Nest -
Nest Hub Max — Everything that Google home hub(Renamed to Nest Hub) had plus the embedded camera. Face match using on-device machine learning provides personalized content.
5.b) Pixel -
Google pixel 3a and 3a XL starting at $399 targeting the non-premium customer segment. Google maps in pixel get the preview of the new directions module using VPS and AR.
6) Security and privacy -
Auto delete controls let you delete the data that was collected automatically after a specific time. Currently, it is available for Google search. Coming to products like chrome, assistant, YouTube, news and maps.
Incognito mode for chrome, YouTube, maps, and search.
7) Language support and announcements -
7.a) Mobile -
Foldables, 5G, On device learning, Smart reply, Dark theme, Privacy settings, Reminders for location usage, Location sharing only when the app is in the foreground, Security patch updates in the background without restarting the phone, Focus mode, Parental control settings are some of the key features of Android ‘Q’
Kotlin first approach for native Android. Jetpack API’ will be introduced first in Kotlin. However, the support will be continued for Java, C++.
Series of educational events on Kotlin can be found at https://events.withgoogle.com/kotlin-everywhere/
7.b) Web -
Flutter gets support for web. The same mobile code base can be used for the web.
Google’s open source JavaScript engine V8 parses JavaScript 2x times faster with 20% less memory usage.
Image lazy loading for chrome by simply adding loading=”lazy” attribute in the image tag.
Lighthouse gets direct support of performance budgets that helps optimize the performance of the sites.
https://web.dev is launched to help understand the modern web.
Chromebooks come with out of the box Linux support.
What do you think was the major theme of Google IO this year?
Happy learning!
Originally published at http://shankarkumarasamy.blog on May 10, 2019.
