Emerging Tech: Personal Calendar Assistant AI
In analysis / By Gina Hall / 26 September 2016
Guess how much of your precious time is wasted getting a meeting on your calendar.
On average, scheduling a meeting involves eight emails, which you have to track over 2.5 days. All in all, it takes about 15 to 17 minutes to put on the books that coffee chat with a new client or the call with your supplier. And just think about how many of those meetings get cancelled, starting the process all over again.
Enter Amy Ingram, a new virtual assistant poised to take the business world by storm.
As assistants become a luxury few in the C-suite can afford, Amy is a bot that promises to save an executive’s time by handling the back and forth of scheduling a meeting.
Say you connect with a new contact at a conference, who follows up for a lunch meeting. You confirm that you’d like to meet up, then all you have to do is cc Amy in on the conversation.
Amy, who has access to your calendar, will handle the emails with the new contact to agree on a time and location that works best.
The bot is in development at the New York City-based startup x.ai, which was infused with $23 million in April to bring Amy to market later this year. The company has put lot of thought into developing the personality of Amy -- or Andrew, if you prefer a male assistant -- to ensure that your contacts feel like they’re interacting with a coherent consciousness.
The more you interact with Amy, the more the bot will pick up on your meeting habits, such as your preferred coffee shop or pub.
“Amy is modeled on your interactions with a human assistant,” Stefanie Syman tells the Nasstarian. “Many people mistake Amy and Andrew for actual humans. It’s very magical.”
But, of course, it’s not magic, it’s artificial intelligence (AI).
While everyone is talking about AI, there’s a lot of confusion in the market about how to integrate it into the workplace. What x.ai has done with Amy is attacked a very time-consuming chore, but one with a limited scope which a machine can understand.
So how does it work?
Think about it like this, there are three core parts of a meeting: time, location and people. The machine extracts that data from email exchanges and labels it. The company also has human AI “trainers,” who oversee how the machine labeled that data.
“We’ve gotten Amy and Andrew to a pretty mature and smart place,” says Syman. “We remain hungry for data. We’re still training the machine quite a bit. As we add new skills, we need to train on a new data set and continue to collect massive amounts of data.”
Amy and Andrew are telling some interesting stories about our work habits, too. Not watercooler gossip, mind you, but how people behave within workplace cultures and how they differ from region to region and from role to role.
Soon, x.ai may have enough data to pinpoint the best watering hole or coffee spots for meetings in your area. The company has already been able to determine how far in advance Americans plan as opposed to the British. The British schedule further out, while Americans are last-minute planners.
“Eventually, we’ll be able to do some really interesting analysis,” says Syman.
Amy and Andrew don’t always comprehend the nuance of human language and will sometimes defer to you to hash out more intricate conversations. The bots currently handle tens of thousands of meetings per month during beta, but they will likely have to surpass the accuracy of their human counterparts to gain a foothold in the marketplace.
“People are far less forgiving of machines,” Syman observes. “Andrew could make a mistake that is easily corrected and gets the meeting right in the end and not be forgiven for it as quickly or as completely as a human assistant who made the same mistake.”
The company is coming to market with a professional version of Amy this year, with an edition for teams and small businesses in early 2017. That’s probably not soon enough for the 90 million knowledge workers and who schedule about 10 billion formal meetings a year just in the U.S.
“We want to schedule every single one of them,” says Syman.