US start-up datalog.ai introduces AI that talks our language

US start-up datalog.ai introduces AI that talks our language

US start-up datalog.ai introduces AI that talks our language
US start-up datalog.ai introduces AI that talks our language

San Francisco-based start-up, datalog.ai, has introduced an artificial intelligence (AI) platform that supports continuous conversation with a bot via natural language understanding.

The platform, known as MyPolly, allows developers, users, businesses or consumers to build their own chatbot that will not just answer ‘yes’ or ‘no’ questions, but will learn from and respond naturally to human conversation.

The product is still in closed beta testing for developers and bot builders, but co-founder and CEO of datalog.ai, Jack Crawford, told Internet of Business the company plans to launch in the next four to six weeks.

Related: UK’s NHS trials AI app as alternative to medical helpline

datalog.ai making AI more human

In a phone conversation with IoB, Crawford explained that the technology is built on deep learning capabilities, with the focus on creating a system that learns as it works with an individual or consumer.

“What this is, is the ability to carry a conversation across the course of using a bot or a virtual assistant,” he said. “So our initial market is to work with bot builders so they can get this out to market very quickly and make the bot three things: more intelligent, more personalized and conversational.”

This is about making an AI that assistant that is more personal than an Amazon Echo or Google Home. The only comparable product we’ve come across is Olly, a social assistant which designed to develop a different personality for each user in a household.

But that’s not what Crawford and co are trying to achieve with MyPolly, and it does not possess natural language understanding, so the comparison stops there.

Man chatting with chat bot on smartphone vector illustration. Communication with chat bot use smartphone, message for chat bot

Opportunities on both sides

This product aims to create a better experience for end-users, but also to generate revenue for business.

One possible use for MyPolly lies in the Amazon Alexa skill set, which as we’ve noted are somewhat limited at present. Crawford pointed out that no one has created a conversational skill yet, so there’s an opportunity to do so, and it’s these kind of partnerships that will drive MyPolly forward.

Using a hypothetical example, Crawford notes that if a consumer goes to an H&M store, which uses a bot via Facebook Messenger, they may get asked questions such as ‘are you male or female?’, or ‘Are you looking for dresses or coats?’

“This isn’t a conversation. What happens is you’re taking down a menu and the menu’s not a bad one, it helps you get to the point where you might find the product and actually go straight to a shop and buy it, but what our software allows you to do is have that person have a conversation just like you and I are having but keep track of what’s going on.”

On the business side, Crawford gave the example of bots replacing human workers at a call center. It may cost $10 to hire a human worker to take calls in America or $1 in India, but “if you don’t have a call at all you’re saving a ridiculous amount of money.”

“If the consumer has a better experience and they’re not on hold and going through menus but actually having a conversation, that’s the win in both ways: customer experience and lower cost in solving their (the customer’s) problem.”

Related: Lost jobs but happier customers — insurance embraces AI and IoT

A subscription model

So how does datalog.ai make money?

Crawford told IoB that there won’t be a device for MyPolly, “so our goal is to let others build software…we’re a technology platform.”

“Our revenue model is if it works for you and you make money out of it then you share part of that through a subscription or licensing fee.”

The company will offer its platform for free along with a “fairly large amount of transactions” – meaning an interaction between a user and a bot – and there are two reasons for this.

“One is that we want to make it available to everyone. The second is that you need to use it to help train it. We know you can’t benefit to make a better product unless other people use it.”

It’s a smart model, and with roughly 900 million people using Facebook Messenger each month alone it’s easy to see how MyPolly might profit in this space, as bots like Cleo and Fynd have done.

We’ve seen a whole host of AI assistants released in the last 6 months, but the introduction of MyPolly will surely make the space that little bit more competitive.