Maluuba

Content sourced from Wikipedia, licensed under CC BY-SA 3.0.

Maluuba Inc. was a Canadian AI company focused on language understanding. It started in 2011 in Waterloo, Ontario, founded by four University of Waterloo students: Zhiyuan Wu, Joshua Pantony, Sam Pasupalak and Kaheer Suleman. They began with a voice-based flight search and soon grew into a research-focused company.

In 2012, Maluuba raised $2 million in seed funding from Samsung Ventures. The company partnered with smartphone brands, smart TV makers, automotive companies and other IoT players to bring its language technology to devices. In 2015, Maluuba raised $9 million in Series A funding from Nautilus Ventures and Emerllion Capital and opened an R&D lab in Montreal later that year. The Montreal lab was guided by Yoshua Bengio and Richard Sutton and worked with the Université de Montréal’s MILA and McGill University.

By 2016, Maluuba had over 50 employees and published about 15 research papers on language understanding. The team demonstrated machine reading abilities, including a system that could answer questions about Harry Potter. They also introduced EpiReader, a reader that outperformed Facebook and Google on machine-comprehension tests using large datasets like CNN/Daily Mail and the Children’s Book Test. The company released two datasets, NewsQA and Frames, and conducted extensive work on dialogue systems, natural language understanding, state tracking and generation. They published papers on learning dialogue policies with deep reinforcement learning and freely released the Frames dataset in 2016.

In January 2017, Microsoft announced it would acquire Maluuba for about $140 million. After the deal, some Waterloo staff moved to Montreal as the company’s research continued there. Maluuba’s technology has been used in consumer electronics on more than 50 million devices, enabling smartphone voice assistants and other automotive and IoT applications.