With Amazon’s Alexa, Apple’s Siri, and countless other bots helping us organize our lives on a daily basis, the opportunities for artificial intelligence to expand across all aspects of our daily lives are huge. In a recent study, nearly half of small businesses surveyed said that they are ready to introduce bots and AI into their operating model.
These smart technologies are emerging even in verticals and industries you might not expect, like accounting. Earlier this year, software solutions company Sage introduced Pegg, the world’s first accounting chatbot. Pegg serves as a smart assistant to help companies track expenses and manage finances through popular messaging apps like Slack and Facebook Messenger.
“I hope we will be able to break from Hollywood-born stereotypes and find robots being more useful in our daily life,” says Kriti Sharma, VP of Bots and Artificial Intelligence at Sage. Sharma is responsible for Pegg and other AI initiatives within Sage. Pegg has more than 20,000 early adopters in 110 countries and can manage expenses, financials and future projections.
During this year’s Sage Summit, Sharma talked to Hype about her unique position at the company (she made up her own job title!), why Pegg is one of her favorite projects to date, and how AI can improve our lives for the better.
Tell me more about your current position at Sage.
I am the VP of Bots and Artificial Intelligence, which is a title I pretty much made up for myself. It shows the importance that we at Sage are giving to AI, because we really believe in the power of automation. AI is not taking over human roles, but augmenting and supporting humans to deliver better and be more high performance, highly productive using the technology. That’s what I focus on.
What current project are you most excited about?
There’s quite a lot going on in our world, where we’re building awesome technologies, some of which you’re seeing already. The key one that’s already out in the market is the world’s first accounting assistant called Pegg. We launched it last year, and it works as an interactive response personal assistant in your Facebook messenger, on Slack, or Skype. It just works like a friend — it takes care of all of your finances and business admits. If you want to create an expense, you just send it a message, “Hey Pegg, I spent ten dollars going to X.” It just takes that information, pushes it back into the expense system, and there you go.
It chases people on your behalf, has them pay you as they owe you invoices and payments, but also starts to give you advance analytics-related information on how your business is doing compared to others. And all of this just through natural language, just like you’re talking to me… as a person. We’re building it globally in 23 different countries, so it is pretty big.
Pegg also has a machine learning component, correct?
Absolutely. It has a natural language understanding, machine learning, and we’re doing some experiments with deep learning also. A lot of focus on voice technology is coming up in what we showcase now. Our latest demo is just voice-enabled interactions, which we believe is the next user interface where even typing and sending a quick message can be just too much work.
How secure is Pegg as an assistant and how can entrepreneurs be at peace when using it?
That’s a great question. The way we think about bots and AI is pretty much the way you design any other system using websites or mobile apps. I think a lot of it will relate to consumer confidence and educating people on how to interact with AIs. This is yet another user interface.
In the mobile app world you’ll press buttons and click on forms and drop downs. In the bot world, we use the same information using words. So it is not very different. But we believe there are some design principles bots and AI developers need to follow to communicate that message effectively. For example, a bot should not pretend to be human, and humans should not have access to interactions with bot data.
We also pay a lot of attention to which use cases we allow on what channel or medium. For example, you might have friends coming over and visiting you, you don’t want them to just say, “Hey Pegg, what’s my balance?” and your friends can actually get information which would be private and sensitive. Like with any other technology, you always have to design for specific uses. We believe that it’s voice, you have to be careful about other people listening around you. So it’s not a tech problem, it is more a human interaction and social problem. You might ask for voice biometrics, and it might try to recognize me from my voice signature, or there’s a special code that you have to emphasize.
Why is great design so essential for effective AI?
It’s very important to find the right fit. For example, some mobile companies make the mistake of moving their websites to mobile apps, which is taking the big website and squeezing it into a smaller screen and giving it to users. Those apps fail because users love an app that just does one or two things very well, instead of having too many buttons. The same is true for AI —you can’t just take whatever is in mobile and make it a conversation, you have to design for conversations. It’s not easy to spot it for the first time you see these technologies, but it’s a lot of focus on design in AI, but this is not just about user interface or screens and beautiful font and colors. This is about words. Every word the AI says is beautifully crafted, and it’s a lot of focus as well.
