If you’ve read a book or an article on artificial intelligence there is a very good chance that you have heard of the company X.ai or its founder, Dennis Mortensen. X.ai is a personal assistant that automatically books your meetings, eliminating all the grunt work of connecting and rescheduling. It’s a phenomenal piece of technology which is laying the foundation of tomorrows reality.
- 01:50 Dennis Mortensen’s Background
- 04:18 Predicting technology trends and building businesses around them
- 05:46 X.ai and Solving a Real Business Problem
- 07:48 The Biggest Challenge building Amy (X.ai)
- 12:35 The Future of AI in Marketing
- 13:45 Example of How AI is Changing the User Interface (UI)
- 14:35 The Paradigm Shift – The World is Changing
- 15:25 Advise for Businesses Looking to Implement AI
- 17:15 Who is Using Amy and Andrew (Intelligent Agent from X.ai)
- It is easy to be average at seven things, and almost impossible to be world class at one. The ‘secret’ is to be hyper-focused and do one thing very well.
- I am a firm believer that there are hundreds of thousands, potentially even millions of mobile apps and desktop applications that don’t belong in their current paradigm. There’s going to be a massive transition from past application paradigms into this new conversational paradigm. In this shift, I think we need to look at all the existing applications and all the existing tactics, and ask ourselves – “do they belong here?”.
- When it comes to marketing we need lean back and honestly evaluate whether what we did yesterday is something that’s going to work tomorrow?
- Try to make a distinction between Conversational UI and the Intelligent Agent, because the UI can be tested cheaply and if successful, you can then figure out how to implement this as a machine agent.
Justin Jones: If you’ve read a book or an article on artificial intelligence there is a very good chance that you have heard of the company X.ai or its founder, Dennis Mortensen. X.ai is a personal assistant that automatically books your meetings, eliminating all the grunt work of connecting and rescheduling. It’s a phenomenal piece of technology which is laying the foundation of tomorrows reality.
Dennis has built and sold multiple companies to organizations like Yahoo and Out Brain. He has successfully predicted technology trends and then built businesses around them. Although X.ai has only been around since 2014, it’s already a technology darling with investment giants like Softbank Capital backing Dennis’ company and vision.
Dennis, what inspired your move into artificial intelligence and the creation of the ultimate personal assistant.
Dennis Mortensen: I would like to believe that there’s some sort of coherent narrative in my path up until this point. Back in the mid 90’s we set up shop and focused on log file analysis. We had web servers spit out raw files and we would try to extract insights from that in a very manual process. I had the very fortunate outcome of selling that company in April 2002 – getting out on the right side of the dotcom boom.
After that we worked on creating Index Tools. This tool allowed the customer to extract the insights themselves. Basically, put java script tags on a page, we collected and stored all that data, so you could retrieve that from a web interface yourself. We now know this is Google Analytics.
Thereafter we set up a company called Visual Revenue. Not only did we allow the customer to extract the insight himself, but we provided a set of recommendations where insights were automatically extracted.
Then lastly, if you look at what we do today with Amy at X.ai, we’re building upon our previous experience and insights. We’re not only collecting, storing and retrieving data for insights and recommendations – we are actually taking action. If you ask Amy (AI Assistant) to set up a meeting, she won’t give you insights into how you can do that. She will do the job for you.
I think there is common thread between us trying to create value from data 20 years ago, up until this point. We are still trying to create value from data, except with more sophisticated technologies.
Justin Jones: It’s impressive how you’ve managed to predict and capitalize upon technology trends over the years. When did you start building X.ai?
Dennis Mortensen: We brought the band back together in late 2013, wanting to figure out whether our idea was even plausible. Then we set up shop in April 2014 with a little bit of seed money (about two million) to come up with a data set we could use to test our hypothesis. This was not your typical seed round with a minimum viable product (MVP) etc.
We’ve always known that this journey would be much more dramatic than your traditional weekend hackathon, or your three months in Palo Alto with a startup incubator. We’ve been working on it for three years and have just recently started to commercialize it.
Justin Jones: Back in 2014, I would imagine the challenges would’ve been a lot harder than had you started today?
Dennis Mortensen: For us it wasn’t about becoming an AI company. It was the idea that if you looked out the window and picked the first 20 people walking on the other side of the road and ask, “have you set up meetings this week?” They’ll all say “yes”. Then ask, “Did you like that experience?”. Everyone hated the experience, and that was the pain we focused on.
It seemed that the existing solutions and those over the last decade … were not what the market wanted. What they really wanted was a human assistant that could sit in the front office and take care of requests. They could say, “Hey Tom, could you set up a meeting with Dennis the first week of July when I’m in Manhattan?” And then it is Tom’s job to make it happen.
The sad thing is that Tom will cost about $60,000 a year, so most people couldn’t afford it. So, it falls back on you to set up the meeting. That suggested to us that the market didn’t want a new solution, but rather an affordable replica of Tom. The only way to solve this market need (building a Tom replica) was to create AI. We believed that Amy and Andrew (Intelligent Agent) could solve this problem in an elegant way.
Justin Jones: What was the most challenging piece in designing and creating Amy.
Dennis Mortensen: In hindsight, it’s always easy to figure out what you should have done two years ago, but somehow, we couldn’t see it back then. If I could it over we wouldn’t focused more on defining the universe (everything involved in scheduling meetings) – we should have invested more time and energy into this early stage work thus minimizing the need for future re-engineering.
Clearly defining your universe with a set of annotation guidelines where somebody else can sit and label this data aligning with your definition. If you can’t do this, then you don’t have a data set. At this point, it doesn’t matter how many PhD’s you have on staff because they can’t build any models, they cannot replicate or predict intent. So, setting up annotation guidelines, as boring as it sounds is essential to success. When I look back over two and a half years, I wished had invested way more.
Justin Jones: Are you planning to build other services around your core AI offering?
Dennis Mortensen: We’re hyper-focused on making our two agents, Amy and Andrew successful. It is easy to be half-assed at seven things, and almost impossible to be world class at one. So we’re trying very hard not to dream up fifteen other things. It’s easy to go to the White Board and come up with 15 good ideas related to our core solution. But we don’t. We’re just a hundred people in New York trying to make the most of Amy and Andrew and their ability schedule meetings. That doesn’t mean that we lost all ambition. What you’ll see is X.ai expand on three dimensions.
Firstly, we will expand into other languages. Not just so we can increase the potential market, but also enhance the system so it becomes ‘superhuman’. Imagine being able to hire a human assistant that could speak 17 languages? Your assistant could communicate with your guests in their native tongue.
We also want to provide the opportunity for Amy and Andrew to exist in other communication channels. So when you’re talking about meetings, whether that be internally on Slack (slack.com), whether that be externally in an email, or intimately on text with your wife and kids – wherever that might be, you should be able to wake up the AI Agent and say “Get Dennis and I together this afternoon”.
The third dimension upon which we’re going to expand, is the set of connections that we can make on the event itself. For example, if you and I meet up next Wednesday at 2:00 p.m. for lunch, why do I have to log into Open Table (opentable.com) and make a reservation? My AI Agent (Amy) should do that on my behalf. If we meet uptown, why am I booking Uber? Amy should do that for me. If we’re meeting for the first time and I couldn’t exactly remember who you were, why isn’t there a small bio in the invite or a LinkedIn link to refresh my memory?
So more languages, more communication channels, and extensions into elements that naturally belong in the scheduling events.
Justin Jones: Let’s look further into the future. Where do you see Artificial Intelligence (AI) evolving in terms of business and marketing?
Dennis Mortensen: I am a firm believer that there are hundreds of thousands, potentially even millions of mobile apps and desktop applications that don’t belong in their current paradigm.
There are certainly some applications that belong in the command line where many of my engineers operate. That’s exactly how they want to access these applications. There are also applications that exist on the desktop that designers love to use with their big 27-inch monitor, for example Photoshop. There are certainly mobile apps that I want on my phone, like Google Maps to guide me through the city.
But there are also many other settings where the paradigm is wrong. Let me give you an example to help conceptualize the thinking.
You come visit me and stay at the Hilton Hotel. It’s 11:00 p.m. and you want a Diet Coke. Your first thought is not … “hey, let me go to the AppStore, let me find the Hilton Hotel App, download and install it. Let me set up an account, let me log in. Let me find a local retailer and add Diet Coke to the basket. Let me check out. Let me get my credit card details.” You would die of thirst before you got the Diet Coke.
That’s not what you want. What you really want, is perhaps a number on the door that you can text to the hotel lobby and say, “can I have a Diet Coke please?” Or there’s an Alexa (Amazon) device in the room and you just shout out loud, “Diet Coke please”, and it arrives.
I think there’s going to be a massive transition from past application paradigms into this new conversational paradigm. In this shift, I think we need to look at all the existing applications and all the existing tactics, and ask ourselves; “do they belong here?”.
With regard marketing in particular, there might be many tactics and processes that just don’t belong where we’ve put them. But somehow, we didn’t have any other option when first created or installed. However, I do think that with AI, and Conversational UI (User Interface) in particular – new options are being made available, giving us an opportunity to re-evaluate those past choices and engineer more efficient and effective solutions.
When it comes to marketing we need lean back and honestly evaluate whether what we did yesterday is something that’s going to work tomorrow?
Justin Jones: In terms of businesses exploring the potential implementation of AI solutions. What advice would you give a business or marking agency?
Dennis Mortensen: I think it is important to make a clear distinction between Conversational UI and the Intelligent Agent. The Conversation UI is a new UI paradigm which does not consist of pixels and buttons. Essentially allowing you the ability to interact with customers without the need for intensive engineering.
I think there are many settings where Conversational UI would add a lot of value. For example, I don’t want to go through a long FAQ to figure what how and when to upgrade my phone on AT&T. Wouldn’t it be lovely if I could just ask an Intelligent Agent, ” Can I upgrade my phone on November 1st?” Then all AT&T need to determine is, what did he ask, and then figure out how to answer the question. For all intents and purposes, you could just hire a human to answer the question. But if you established that the customers liked this way of communicating, and like this User Interface (UI) in particular then all you have to do is figure out how to make that human; machine based. So that’s certainly a whole lot more cost effective.
Coming full circle. Try to make a distinction between the Conversational UI and the Intelligent agent, because the UI can be tested cheaply and if successful, you can then figure out how to implement this as a machine agent.
Justin Jones: Dennis, who is your ideal customer. Who are you introducing Amy and Andrew to?
Dennis Mortensen: If I was bold, I think my answer would be everybody. In the United States alone, we see about 10 billion formal meetings being set up every year. These meetings are being set up by approximately 90 million U.S. knowledge workers. This assumes that every single person who walks into an office and touches a computer in some way, will likely set up a meeting. Some will set up many and others only a few. But they all set up meetings.
Conceptually, we want to help all of them. That is perhaps too grand for a small startup like ours (100 employees). So, we are initially going after individuals in pain. For example, a recruiter at Netflix, or a sales manager at the New York Times, people at a startup, or a CEO and founder that simply can’t afford having an assistant at $60,000 per year. They go to our website and realize that they can buy an assistant for thirty-nine dollars a month – they sign up and they’re happy.
That’s one segment. The second segment; we’ve just launched our business edition. Now you’re able to sign up large teams where every member gets their own assistant. This would be crazy if you had to hire humans – but becomes very economical and efficient when leveraging intelligence machines.
That’s where we’re starting. However, it won’t hold me away from my grand vision of helping everybody. I’m sure you can see the awesome output we could get when everyone is using this technology. For example, for our meeting at some point I said I’d be happy to have this conversation. I’ll copy Amy (Intelligent Agent) and she can put something on my calendar. You say “wonderful, I will copy Andrew (Intelligent Agent) on my end”. At which point, Andrew and Amy (both X.ai) can hash out the details and set the appointment.
This is a magical moment, which we refer to as scheduling nirvana. As Amy and Andrew are in fact the same Intelligent Agent (both X.ai), the scheduling response is immediate and with very high accuracy. This is what we’re trying to achieve – this is where we begin to see a real network effect.
x.ai – is a personal assistant who schedules meetings for you