Google betting on AI from India

By:  Peerzada Abrar

Firms that help medicos, travellers, cooks with insights are high on the agenda

It’s no coincidence that at about the time that Google CEO Sundar Pichai has professed the company was moving into an ‘artificial intelligence first’ world, the Internet giant is backing AI start-ups in India to help it achieve that goal. Young ventures that analyse medical data to reduce human error, gather insights to help travellers and provide cooking apps for different culinary needs, are some of the AI companies that Google is nurturing at its ‘Launchpad Accelerator.’

“The pace at which you can solve problems is so much faster when you apply (AI),” said Paul Ravindranath G, program manager, Google India. “We have begun including companies that are meaningfully solving problems… using AI and machine learning.”

The firm’s six-month accelerator programme matches young companies from emerging ecosystems with the best of its people, network and advanced technologies to help scale their products. It also connects them with mentors from top tech firms and venture capitalists in the Silicon Valley, U.S.

In return, Google, whose parent company Alphabet Inc. reported a revenue of $90.27 billion in 2016, is hoping that its technology and cloud computing platform gets consumed by these start-ups and they build their innovation on top of them.

One such firm is SigTuple which is building intelligent screening solutions to aid diagnosis through AI-powered analysis of visual medical data. The firm aims to improve speed, accuracy and consistency of a number of screening processes. Also, medical institutions can serve more patients, with a significant reduction in human errors.

When the Bengaluru-based firm wanted to expand in the U.S., Google mentored the company on fulfilling regulatory requirements such as getting the U.S. Food and Drug Administration approval.Google also helped the firm identify manufacturing companies in the U.S. when it found the process challenging in India. “They try to identify areas where you need help,” said Rohit Kumar Pandey, 34, co-founder of SigTuple.

AI chef

Through its accelerator programme, Google is also rigorously fostering TensorFlow, a software that makes it easier to build AI systems. This is being used by Kochi-based start-up Agrima Infotech. It had built a deep-learning computer vision technology which could be used for advanced image recognition techniques. Agrima has implemented this in ‘Recipe Book’ a popular cooking app on Android Play store. The app solves the recipe discovery problem. For example, a user needs to just click an image of an ingredient. It then quickly recognises and suggests the best recipes which could be made out of that ingredient.

Anoop Balakrishnan, 28, CEO of Agrima, said the ‘good thing’ about Google’s acceleration programme is that “it doesn’t take any stake” in the companies. ‘Recipe Book’ saw more than 2.5 million downloads in its first year and has spread across 104 countries, according to the firm.

Google’s accelerator programme is also backing Noida-based RailYatri, whose app helps long-distance travellers at every stage of the journey. This includes enquiry about seats, the best route to take, information about train delays as well ordering the food and booking the hotel. The firm said close to 40 million people in India take long-distance journeys daily. However, the transportation system used by these people are constrained by supply-demand mismatch and operational inefficiencies. Manish Rathi, co-founder of RailYatri said its platform efficiently uses crowdsourcing, data analytics and intelligence to help travellers make smarter decisions.

New tools, old problems

Google was among the earliest to start buying private firms to advance its own AI research and is now the most active acquirer in the space, according to CB Insights, a data intelligence platform. This July, Google acquired Bengaluru-based Halli Labs that applies modern AI and machine-learning techniques to solve old problems.

This month, dunzo, a task-fulfilment start-up that leverages AI and human operators also raised a $12.3-million Series B round led by Google, according to Aspada, an investor in the Bengaluru-based firm. Dunzo provides a conversational mobile commerce platform. It manages the discovery and fulfilment of local tasks across categories such as food and grocery delivery, and offline retail purchases.

In 2013, Google also picked up deep learning and neural network startup DNNresearch from the computer science department at the University of Toronto, said CB Insights. It said this acquisition reportedly helped Google make major upgrades to its image search feature. In 2014 Google acquired British company DeepMind Technologies for some $600 million.