Meet the Australian startup using artificial intelligence to help blind people see

The original concept for Aipoly emerged from a homework assignment given at Singularity University– of which Cheng was the only Australian among a cohort of 80 students.

The assignment was a lofty one: Create something that would impact the lives of a billion people within the next 10 years.

“The vehicle that we chose to do that through was starting a company,” Cheng says.

After meeting with a Google executive who described her vision impaired friend’s day-to-day struggles, Cheng and her two co-founders had the idea to combine an image- recognition algorithm with a smartphone to give instant feedback on surroundings.

With an MVP in hand after only a few days, Cheng began to meet with numerous blind organisations to get feedback.

“They were a bit jaded and when we first started talking to them they’d say, ‘we’ve seen all these technologies before and they haven’t actually made as much of a big impact as we would have liked’,” she says.

“But when we showed them the technology they were absolutely blown away. The responses we got were really amazing – we were surprised about how positively we were received.”

Buoyed by this feedback, the team threw together a video describing the product which was soon picked up by major publications around the world – and the startup was just a week old.

“We decided to go ahead and turn Aipoly into a bigger thing,” Cheng says.

From MVP to beta

The trio then bunkered down to work on a full prototype of the technology, soon creating a product that could identify 800 objects and thousands of colours.

While the MVP required users to take a picture of an object and send it to the Aipoly servers, the newly-created beta version completed this in near-real time, without the need for an internet connection.

“With our new technology we reduced that time down to a third of a second,” Cheng says.

“Someone can use their phone and recognise three objects in a second.”

The Aipoly algorithm can also be trained by users to recognise new objects that it doesn’t know.

The beta version was released three weeks ago on the eve of one of the biggest tech conferences in the world.

“We were very, very relieved it actually made it up there,” she says.

Aipoly proved to be a central player in the CES conference’s push to showcase more socially-conscious startups after previously being criticised for a lack of attention on this sector.

Startups for the social good

According to Cheng, working for the social good makes business and societal sense.

“If you’re building something that has good social outcomes then it’s good for business too,” she says.

“People want what you’re doing and will fund what you’re doing.”

It also ensures that something good comes out of the business even if it ultimately fails.

“If your startup is doing something that has a really good social cause then even if you fail, you don’t really fail,” Cheng says.

“Even if you just advance knowledge in that field, or if you create something that helps hundreds of people instead of thousands of people, thousands of people instead of millions of people, at least you made that impact. If you create something that doesn’t help people and you fail, you also failed to help people along the way.

“If you do something that has a social good and you do really well through it then everyone wins. But if you fail or run out of money at least on the way you still changed people’s lives.”

But Aipoly doesn’t look like failing any time soon.

The dedicated team has already released a new and improved version of the app and are setting their sights on a number of new features, including translations in Japanese and Mandarin, and the ability to recognise faces.

With so much success in its short life, Aipoly is well on its way to completing the ambitious homework assignment it was initially created for.

“It’s been three weeks so who knows what other stories we’ll hear,” Cheng says.

Follow StartupSmart on Facebook, Twitter, and LinkedIn.