Or listen in your favorite podcast app
Very few people are more plugged in to the world of A.I. than Rob May (LinkedIn), the CEO of Talla, which is an automation platform designed to completely change the way customer support is done. On this episode of IT Visionaries, Rob and Ian take a deep dive into everything currently going on in the A.I. industry, what should worry people and what they should be excited about. Plus, if you’re interested in learning more about the industry, Rob lists all the resources he uses to stay up to date, and gives some tips to executives who are looking to bring A.I. into their own organizations.
Best Advice: “Get an executive coach. You have to be successful in this job and it is just a relentless job. There’s always so much stuff on your mind. Get some help. Get an executive coach, somebody who can teach you some of the things that you don’t know. And I would encourage you to get one that’s more of a business coach and how to do things that you don’t know how to do rather than just a sort of psychologist who is all about like, you know, hey, how’s life?”
Rob’s history with A.I. — (1:35)
Rob was an electrical engineering major at the University of Kentucky and went on to become a hardware designer. In the early 2000s, he was working on a master’s degree with a focus on artificial intelligence. But at this point, A.I. was still in its early stages, so Rob didn’t seem much of a practical application for the technology. Rather than continuing to research, he went into programming and management but always kept an eye on what was happening in the world of A.I. For the next 10 years or so, Rob worked at a number of different start-ups and found success in technology. It wasn’t until 2014 that he decided to move back into the A.I. field and start his own company. He also started publishing an A.I. newsletter and investing in A.I. companies.
“A.I. at that time was like symbolic logic processing and lists. And so I got really into a lot of the concepts but it seemed to have little practical application. I’m more an applied kind of person. I’m not a researcher. I don’t like that kind of work. And so I just didn’t see any use for this. So I followed the field, but I quit the program and went into sales, engineering and general management and stayed in technology.”
Founding Talla — (5:24)
After selling the company he founded, Backuptify, in 2014, Rob was looking for what to do next. He realized that there was a lot of unstructured text in the world that was about to become computable, and he connected that with the rise of communication software being used in various enterprises between employees and with customers. He saw an opening to use automation to create communication tools and chatbots that actually worked really well and would solve customer service tickets without using as much human capital. This was the genesis of Talla.
One of the initial challenges Rob and the Talla team ran up against was adoption problem. Initially, they were marketing Talla to sales teams, but with salespeople, as long as you hit your numbers, the tools you use don’t matter. So many salespeople continued to do what they always did and few incorporated Talla into their workflow. But, when Talla sold to support teams, they adopted the system in huge numbers. Because support teams were getting the same kinds of help desk tickets over and over, they could automate answers with Talla and greatly reduce the number of tickets they had to deal with personally.
“The adoption pattern was, you sell into a 50-person sales team and six people use it regularly, 10 people use it a little bit, nobody else uses it. You sell into a 60-person support team, 58 of them use it regularly like they’re supposed to. So that’s how we landed and sort of migrated into really focusing on support teams because they had the best use case of having a problem that we could solve with unstructured text and they would actually adopt the product.”
“We can say, look, in our best cases, 85, 90% of the time, Talla can give the very specific right answer, right on time. Just the snippet that you need. It’s not search results. It’s not a list of possible answers. It’s just the one thing that you need. So that’s a better user experience. Then on the other side, you can deflect some tickets from coming into the reps. Reps have to answer tickets, but they can use Talla and answer them more quickly with Talla, which has suggested answers and stuff like that. And so it’s multiple value props we sell in different ways to different kind of customers, but overall it actually provides a lot of different value to the organization.”
The RPA Industry — (15:00)
Robotic Process Automation (RPA) has been around since approximately 2014, but only in recent years has there been growth in that area. Many Fortune 500 companies are bringing in partners and other organizations that focus solely on RPA in order to make their companies more efficient. The problem right now is variability. Variability makes automation harder, so companies like Talla are coming in with new approaches and methods of automation that take variability into account.
“It’s mostly desktop systems and it’s mostly back-office processes. So an analyst comes in, they matched your process, you’re moving data around, you’re opening different systems, you’re scanning invoices and you know, paying bills and stuff like that, and they write a script for all that. And it might cost you $50,000 a year for the script, but then you can run it, you know, repeatedly and it saves you monotonous work.”
Surprises and worries from the A.I. industry — (16:45)
Rob has seen a lot of things coming where it concerns A.I. For example, various fake news and A.I.-powered videos were things Rob touched on in his newsletter a few years back that people thought he was crazy to bring up. What has surprised him are the discussions that take place around the ethics of A.I. and how to figure out real ways that A.I. can be used against us, rather than throwing out wild scenarios that will almost never happen. Rob worries that A.I. can eventually become a rewarding mechanism for some but not all. He gives college admissions as an example, saying that the college admissions process has essentially been automated in a way that certain people with very specific criteria (be it grades, extracurricular activities, etc.) are rewarded with admission regardless of whether they are actually a good fit for the school or not. People have learned how to game the algorithm and Rob worries that as more processes become automated, only those who can game the system in similar ways will be able to reap certain benefits.
“One of the things that surprises me is I think when people talk about A.I. ethics, they talk about a lot of the wrong things. You always hear this example of like, ‘Oh my God, what is a self-driving car going to do when it’s forced to decide between killing a passenger and killing a bystander on the side of the road or killing the driver?’ But one of the things you have to think about with where A.I. is going is you have to think about what do people do, right? This is ultimately about trying to build machines that do things that people do. And how many times does a person make that decision? Never, right? So I’d be much more concerned about algorithms manipulating us in similarly nefarious but maybe not as deadly types of ways.”
Can A.I. create something novel? — (33:00)
Today’s A.I. is built on information and data from the past to create something predictive for the future. But can it be used to create something totally unique? Rob thinks so, and he points to the recent investment in neural networks and deep learning technologies. He believes that A.I. can be used to take those technologies even further.
“In 2018, a lot of people started to realize some of the limits of deep learning. There’s a big debate going on in A.I. asking can it get us to the next level of actual reasoning? And are you going to need something better?”
Investing in the right things — (38:30)
Rob says that everyone always wants to invest in winners. They want to put money on innovation but they want the innovation to be predictable. Unfortunately, that’s not how innovation works. You have to keep trying, keep investing and play the numbers game knowing that, eventually, something might hit, but you’ll take losses along the way.
“Everybody wants innovation and everybody wants to push it… And what they really want is they want predictable innovation. They want to innovate and know that it’s going to be successful. And I don’t think anybody’s figured that out yet. Innovation is a numbers game, right? You have to try 10 things.”
What to be excited about — (45:00)
For those companies and CIOs looking to make A.I. a part of what they do, Rob says they need to have a PAC framework. PAC stands for “predict, automate and classify,” and Rob says that’s the best way to start when you’re looking at A.I. tools.