This audacious start-up wants to build a better Android Search than Google
But Cube 26 which is a start-up known more for making smart bulbs intends to change this and that too with a prowess in machine learning and artificial intelligence, an area Google is known to boss. In an interaction with me, Cube 26 founder and CEO Saurav Kumar reveals his audacious attempt at out-Googleing Google at search on Android.
Expertise in Android and AI
Kumar points towards the roots of Cube 26, which unknown to most started out customizing Android for Indian smartphone brands like Micromax, Karbonn and Intex.
“For Micromax we started three years back. We shipped more than 5 million Micromax devices. We started with the Canvas 4.To consumer, it may look like an app, but for us, it was the complete OS customisation. To lockscreen, the launcher, the overall look and feel of the product,” revealed Saurav Kumar.
Pertinently, Kumar points towards a data science driven approach which is rooted deeply on the locality of content, than a dependence on the cloud. The company claims that it has 14 people working on machine learning and it has consultants and partners in IIIT Hyderabad and also the University of California. You don’t hear that kind of chatter from most Indian start-ups.
“Whenever we did these customizations our focus was on data science. And from the very beginning we have focused on data science. The first product was look-away to pause which was on Canvas 4,” claims Kumar.
Their approach towards the search problem is also quite interesting as it is at odds with what Google has achieved or what most of the big players in the AI space have achieved. If there’s a parallel, then it is Apple with what it is doing with local search and AI on iOS 10.
“There was some amount of machine learning happening and some amount of computer vision, eye tracking and face recognition and some level of Android as we were going into the system,” Kumar explained.
“The one area where we can win this race is local. Google search whenever you type something, most of the search engine is happening in the cloud. They send their algorithm back and the data and everything is residing in the cloud,” he specifies.
He repeatedly points towards the example of iOS and says that his new launcher is doing some things similar to spotlight search on the iPhone in terms of core technology and speed.
From a technical perspective, the core Google apps including search are mostly created using Java, which Kumar and many other programmers deem to be slower language as compared to Objective C which is used for iOS apps. The search on the Cube 26 launcher is leveraging C++ which is similar to the speed on iOS.
More importantly, Cube 26 claims to be providing a localized solution. It feels that their solution could be superior in the Indian context as networks haven’t evolved to a level where users can depend on a fast data connection for a seamless experience.
“We believe local search is going to become big especially with the amount of data on your phone is going increase. That’s where we are well placed because we have an Android customization expertise which enables us to navigate the entire Android maze, and we can walk through the network framework and what not,” said Kumar.
Contrastingly, if Google had to deliver this kind of speed, then it would need to make a full copy of the phone on its cloud which would not only be fraught with a multitude of privacy issues, but it would also create a data consumption problem, in a price sensitive market like India.
In simpler terms Cube 26 is saying is that its search can understand what’s there on your phone – be it a contact, a mail, a calendar or even music and give you context without channeling the cloud, all locally. This is something Apple has proclaimed on iOS 10.
So if you get a mail from a certain someone and then that person is not in your contacts, but then you get a phone call from the person, the search/launcher can understand and tell you if it was the same person from the mail. iOS also does something very similar on iOS 10.
This is just one example of what the search is able to do. Its capabilities are quite broad.
A mobile first approach
The company is quietly confident of the type of experience it can cater to the user. Its mobile first approach helps it to build a UI that’s more tenable than say the simple 10 web links that you’d get on a web search within the Chrome browser on Android.
“Google has to focus on their web based consumers and also mobile. But since we don’t have a web presence we are mobile first. We have scrapped and organized the content on our servers. This is global search and deep linking and bringing it to a mobile first UI.”
Cube 26 claims that it has managed to scrape parts of the web and built a robust search for apps, business, contacts, and also has managed to index some apps like Saavn by partnering with them.
“With our search our focus was around finding the most relevant things around the query. It will be a search app, but it will be part of a launcher. The indexing of data is happening locally on your phone,” says Kumar.
At the moment, the company intends to deploy the launcher and the suite of apps on the Google Play store so that it can maximize the amount of users it can reach. It hasn’t made a full Android fork like say a CyanogenMod, a MiUI from Xiaomi because it felt that it installing a ROM on a phone is difficult task for most users, and this way it could partner with companies for a pre-loaded experience and also attract users of existing smartphones in a simpler way. From a technology point of view, it has all the ingredients.
In addition to this, the company is also launching a music player app, a new camera app and also a SMS app which allegedly excels at blocking out spam. For these apps too, the company is channeling its data driven approach and using techniques like machine learning to filter out unwanted elements or detect music that is being played. The company has also launched a new version of its smart bulb which now is brighter and can do more diverse things than its predecessor at the same price.