Video: How to launch your digital human from scratch, with Insta Mortgage

If you’re thinking of pursuing any new technology innovation project, you’ll naturally want to know as much as possible about what lies on the road ahead. That’s true for any conversational AI project you may be considering – from chatbots to digital humans.

Conversational AI projects in banking and financial services are some of the well established, so there’s a lot of experience to learn on in these industries.

In a recent webinar, we invited Shashank Shekhar to chat about his recent digital human project. Shashank is the founder and CEO of Insta Mortgage (at the time of recording named Arcus Lending), and recently launched Rachel, the first digital human for the mortgage industry.

In the video below, among many other things, Shashank explains:

• Why he has invested in digital humans as someone who is hyper-focused on customer experience and service;
• How he’s planning for the development of his digital human’s skills, and what versions 1.0, 2.0 and even 3.0 of Rachel looks like;
• And three steps for getting your digital human project off to the best possible start.

What’s the next step on your conversational AI journey?

Are you ready to get going with your own digital human innovation project? The best place to start – as Shashank and his team did – is with a free trial of UneeQ Creator.

If you have any conversational AI in the form of a chatbot and/or NLP, you can integrate this easily and start speaking to your digital human in no time at all.

If you don’t have a chatbot, you can start small-talking with your digital human in “try mode”, to get a feel for the experience. Or, you can use our simple Google Dialogflow integration. Just sign up for a free Dialogflow account and, like Shashank, you may find yourself writing for, and speaking with, your digital human in a matter of minutes.

A transcript of the conversation

Tyler: Ok, Shashank, how are you, man?

Shashank: I’m good, Tyler, how are you?

Tyler: Super good. Always happy to talk to you. It’s really, really interesting what you’ve got going on with Arcus Lending. Can you just start by telling us a little about Arcus Lending?

Shashank: Sure, of course. Yeah. We started in 2008. As you know, that was probably the best time to start the mortgage company – everybody was jumping on the ship. I said, why not? Let’s do it. But we started in 2008. Long story. I won’t bore you with that. But it’s been, what, about 12 years?

We have now expanded into 20 different states with offices in multiple states. And in 2017 we were named in the INC 500 list of the 500 fastest-growing private companies in America. We hope to make that list again next year, actually,  given where the growth is this year. But that’s where we are. We’re a mortgage company headquartered here in Silicon Valley – I’m doing this call from San Jose – and with multiple locations across the country. That’s what we are.

Tyler: So obviously, you’re a very brave guy. Right, at the height – or  we didn’t know it at the time, but the beginning of the housing market collapse, you were like, let’s do this. You’ve taken another leap of faith with digital humans. What attracted you to the digital human platform?

Shashank: So, I am always kind of curious about finding out what’s the latest in customer experience, one of the things that even though we are still relatively a small company, but even through the years when we were very, very small. One thing I always used to say is that we are in the customer service and education business, we just happen to do mortgages. And so customer centricity has been the focus of Arcus Lending, whether I was a one-person company or two, or like close to 100 people like we are right now. And when David Lykken connected the two of us, one of the things that really interested me about the digital human is how it can take the focus that our organization has about customer centricity and can actually kind of exponentially grow that part of it by having the digital humans interact with our clients.

And so it was an interesting concept. Of course, a lot of us are used to seeing Alexa and Siri in action – not really seeing but hearing and listening to them in action. It was kind of taking that to a different level with with digital humans and saying that you can actually see them in action. And something that I kind of instantly felt – and this may sound crazy, but literally… listen I was probably sold within 30 seconds of looking at it like “I can see a future with something like this for what we are trying to accomplish”.

Tyler: You nailed my next question, which is sort of a value. So, I guess I would like to see, in your opinion as a business owner and a startup guy, an entrepreneur, why is it crucial for businesses to be a part of the next evolution of conversational AI?

Shashank: I think there are some technologies where you either get part of it or get left behind by years. When the iPhone came out and then Samsung tried to catch up and then it’s still been playing the catch-up game, it’s just by the time you are on version 3.0, somebody is on version 1.0 and they’re still trying to catch up. We have seen – I’ll go back to the the adoption of Siri and Alexa has been through the roof. From a kindergartener to, I don’t know, seniors, everybody they are totally fine with talking to them as if they were just part of their lives. And that’s what’s going to happen to something like a digital human. And the fact that: look, the evolution of experience on online platforms have been what we started with, something as basic as contact forms on websites. OK so let’s do a contact form. Then we kind of evolved and we had lead forms. “Give me give me some more information about what you’re trying to do.” We got even acuter by doing dynamical lead forms. “Hey, if you give me this answer then we will give you some other answer.”

And then kind of that evolved into chatbots, saying “let’s give this experience a little bit more back and forth”. And then we had Siri and Alexa, which is not on the web as a platform – of course I can’t implement Siri on Arcuslending.com if I wanted to – but someone like Rachel, which is our digital human project, takes that to a completely different level, where we call it humanizing the mortgage experience.

And it may sound contrary to the fact that she’s a digital human and not really human, but that’s what it is. When we talk to Siri or Alexa, we feel that human connection and that’s what we are trying to do with with the digital human. And I think that’s – business owners, entrepreneur, entrepreneurs who are not adopting something like this, it may sound a little out of the way right now, it may sound as if maybe people will get used to it in two or three years or whatever, but they’re already used to it. They’re already used to talking to them. And this is taking it to the next level of actually seeing someone in action, having that connection go even further. And I would strongly advise someone to take a look at it, and and if they don’t get sold in thirty seconds like I did, maybe they don’t get it. And I don’t know what their understanding of this is, but I absolutely feel, Tyler, that this is the next level of interaction between consumers and the businesses.

“I absolutely feel that this is the next level of interaction between consumers and the businesses.”

Tyler: That’s awesome. You mentioned Rachel, she’s your digital human at Arcus Lending. Can you tell us – like every employee, every person who begins a journey, we all start with a single step, right? There’s a first chapter. What is Rachel currently focused on? And then how do you see her growing? What’s the next couple of steps in her journey?

Shashank: So interesting, because I know you and I kind of jumped on the call the moment I saw this video and I was like, I need to talk to somebody. And I’m so glad that David connected the two of us, and we had a long conversation about it. But it’s crazy that within that few seconds I could actually not just see the 1.0, but I could actually see the 2.0 and 3.0 where I want to go with Rachel on this. And we implemented and went live a couple of weeks back, and initially we wanted to go live testing some very basic concepts – like can she answer some frequently-asked questions that our consumers and users ask all the time? Can she direct them to the right place and say, “I want to rate code”. “Here you go, you can get a rate code here.” “I want to get in touch with the processing team.” “I’ll have the processing team reach out to you.” She’s already capable of doing that.

And that took pretty quickly, because, as you know, we probably went live in a record time – and we can talk about the the entire execution in a few minutes – but that was the concept behind 1.0. We wanted to be quick. We wanted to be fast. We wanted to be first to the market. We really wanted to be the first mortgage company in the world to go live with a digital human.

And that was the initial concept: that even though we are in a hurry, we still want her to be functional; functional enough where she can answer a lot of questions. And that was 1.0. On 2.0, we are already working with the dev team on building a lot of API integrations. A lot of platforms that we already use, we are trying to see if Rachel can talk to those platforms and get a lot more value both to the consumers, our clients, and also to the backend. A lot of our loan officers and processors, they are trying to find information all the time.

Like “what is the guideline? If this person has only been employed for a year, can we give them a loan or not?” Even employee questions like: “Hey, what’s our 401k policy look like? Do you cover insurance?” We are trying to have Rachel talk to all these different knowledge bases that we have, whether that’s on our blog or through API integration to our loan-origination system, and that’s a lot more work than 1.0, of course, where we are trying to just feed her some basic questions and frequently-asked questions. But already, just based on initial conversation with the dev team, it looks like all of that is possible and much more really. So, I’m super stoked already about what in three months’ time Rachel can do for us both for our prospects and clients, and also for our employees.

Tyler: Wow. Wow. So Rachel is starting at Arcus Lending to help customers. But you’ve just mentioned loan officers, people underwriting the loan, looking for information related to the client’s data that they’ve submitted. And you you mentioned sort of a human resources angle where Rachel can just answer questions like, “hey, do we get Columbus Day off?” You know –

Shashank: “No, no you don’t.”

Tyler: Hahaha, Rachel’s answer is just “no” for everything! “How about Christmas and New Years?” “No, I work 24/7, you should too!” No, her job is to, yeah, help people find the information that they’re looking for, in the easiest way that they know how to ask for it, which is verbally.

Shashank: Yeah. So, we started with Rachel with three big, I would say, initiatives – or three big questions, three big asks from Rachel that can she fulfil these three things. The first thing was can improve the user experience; and I said as a customer centric organization, that was the number one thing; and even if she could do just that, I would have been fine with that. But I was like, OK, can she improve the customer experience when they come on the web, can she direct them to the right places? Can she answer mundane questions like what’s the status on my loan, where they don’t have to email the processing team and they can take five hours to get back to them – especially right now when the mortgage industry is going through the busiest year that they’ve ever had on record. Rachel can take a lot of those questions.

But the other two priorities for us, the KPIs so to say, for Rachel was can she improve lead conversions given when users come on the website and talk to Rachel, can we get say, if we have 20% conversion rate now, can that move to 30? Can more people give us information about the fact that they want a mortgage?

And third, which is kind of, as you said, a little out of the ordinary, was actually to help the employees themselves and not just not just our users and clients into getting them the information, saving them the time that they currently get from all over the place or from our HR which, again, may not be an instant answer, right? You email someone and then you wait – or you go through the entire guide of the employee handbook and try to find the answer yourself. So three very different KPIs, but all of which Rachel is capable of as of now, at least, the way we are looking at it.

Tyler: You know, you mentioned a point there that I think is very important to everybody who will be listening. To this broadcast, it’s about increasing the business’s bottom line, right? So what you mentioned was “if lead conversion is 20%, can we get to 30? Hypothetically, what does a 10% increase in lead conversion sort of look like to the business?

“Even for the size of company we are – and, as I said, we are relatively small – you’re potentially talking an extra two or three million dollars in bottom line because of where Rachel can be when 2.0 goes live.”

Shashank: It depends. Like we are, what, about a 15-million-dollar company in terms of revenue, so there you go, a 10% higher lead conversion means an extra million and a half dollars in revenues – right there. And we are a growing company. If you keep multiplying the 10% with higher revenue numbers, then that’s a much higher number. And we talked about the fact – and that’s just on pure revenue from the lead conversions that go higher. But a lot of questions to our processing team right now, backend, is just about status calls. A lot of – I would say about 30, 35% of their time is spent on “hey, what’s the status of my loan, and what’s going on next?” Rachel can handle that. And imagine the productivity and efficiency that team gets. That team can potentially do 35% more work for the same salary that we’re paying them because they have much more bandwidth now to work on that. So if you accumulate all of that, even for the size of company – and as I said we are relatively small – you’re potentially talking an extra two or three million dollars in bottom line because of where Rachel can be when 2.0 goes lives.

Tyler: I think now’s a great time to talk about how you brought Rachel to life. Talk about the construction, if you will, of this project and who you worked with.

Shashank: Yes, I would love to. And in talking about costs and of course, I don’t want to get into exact dollars here, but one of my biggest concerns when I picked up the phone and called you, Tyler, was: this is going to be so obscenely expensive that, even though I’m in love with it, this will be beyond our budget And I was pleasantly surprised by the fact that this came at a cost where really any company can implement it. I know that we are slightly bigger than a startup that probably started six months back, but we are still a very, very small company. And even if you started literally a couple of months back as a company, this is something you can afford. Of course, I encourage you to go to digitalhumans.com and find out the specifics, but if you’re just listening to this and wondering how much it costs, it’s really something that pretty much every business can afford is all I can say.

But let’s deconstruct how we went from having that conversation that Tyler and I had to really going live. This took all of probably six weeks, if I remember; from that conversation to Tyler mentioning how our Creator platform is going live in about a week and a half, I think from that point of time, which is where I would start my clock, because until then it was just a concept and we were talking. But the day that went live, we bought a plan.

And from there on – I think the biggest thing about the Creator platform is that it does the biggest heavy lifting, which is creating a digital human for you; because after that, it’s just just the brains that you are creating, because most of it’s templated – you get different looks, you get to play with it, you can have English and US accent or UK or whatever – so you get all those all those different audio options. All of that is inbuilt, which was crazy because really in about five to seven minutes, your digital human is live. It may not have all the brains that you want it to be, but in terms of you can actually start interacting with it. And I’m not a tech guy, I don’t have an engineering background, but it’s so crazy that after we picked the the persona that we wanted to, Tyler, Piers and I went on a call in about 45 minutes, I was able to program Rachel, which is beyond crazy to think. My engineer friends kind of – I throw shade at them: “look, I can get this live, I can let her do whatever I want her to do.” And then, if you can actually believe it, later that evening, my seven year old was trying to create questions on Google Dialogflow and saying “Dad, what if I ask her that?” And I would go to Dialogflow and create that question and get the answer.

So, that was beyond crazy because that was another thing – me not having a tech background, I would think that this would take a lot of my bandwidth or I’ll have to work with a very expensive dev team to get it live or get it working. But even complicated stuff like, what sounded complicated, like SSML and Google Dialogflow and everything – once they guide you through it, a lot of the setup, you can do it yourself. And that’s that’s the part of it, is that you can take the Creator platform and have a basic version of the digital human go live in probably 24 hours. There’s really not much work that you have to do at that point of time. And of course you can keep building on it like the way we plan to  with Rachel 2.0 and hopefully then 3.0 at some point of time. But from getting on the Creator platform to getting it live was probably 24 hours, but even building the brains and everything else that we wanted her to have (all the knowledge that that we wanted her to have) that was five to six weeks from that point, to actually making a press release about us being the first mortgage company in the world to go live with a digital human.

Tyler: With everything we’ve talked about today, just to kind of bring it all together, in your opinion, what’s the most important area to invest in in order to have a successful digital human employee? You know, what’s your advice to other people who are listening this that might be considering it? Where do they start? You know what I mean, just to help them get going?

“Be very clear in what you want to achieve from this digital human – because the potential is limitless.”

Shashank: Sure. First of all, I can’t even imagine anyone other than, say, UneeQ, being your partner in doing it. They have been true partners. And I don’t mean to say just because I’m having this conversation with Tyler, really every step of the way someone from UneeQ was there with us helping out (with absolutely no background  that I had with this) every single day as when we had a problem, there was someone to help us out. So that’s that’s the first step is – I don’t think any other company can do it better than this. That’s what we say, even on the mortgage side, it’s like if you’re working with a great loan officer then that loan officer will probably handle most of your problems anyway. So that’s your first step, is that you partner up with the right people.

And then once you do that, be very clear in what you want to achieve from this digital human – because the potential is limitless. As I said: we had three KPIs that we wanted to go after in the beginning, and that’s what my focus has been with blinkers on – that “this is what I want Rachel to do”. Similarly, you should be very, very clear on what you want to do; and it might be a good idea to go with one at a time, and saying, “OK, there are three KPIs, let’s work on one, and then we go with two and go with three”. Because it’s not that difficult to train her or him, whatever your human looks like; it’s just that if you start slow and get one step done at a time, it makes you feel that you’ve achieved something and not just stuck in that implementation phase all the time.

I’m huge on execution, Tyler, and that’s why I want to see something go live and then we will work on it after that and we’ll keep adding on it. And while I don’t like that, “go fast and break things”, it’s more than “go fast and make things, but keep making it and then keep making it better”. I think that’s the approach that we should be taking, is that we work with the right partner, you work on one thing at a time and then then move on to the next thing and perfect it. Really, if you do it right, your first version could be live in two to three weeks, and then you work on the more difficult task and the more difficult task. As I see it, this should help us save hundreds and hundreds of hours and hopefully make millions more as the time goes by as Rachel gets better at what she’s doing.

Tyler: Anybody else you want to give a shout-out to, and we wrap up?

Shashank: The people that Spark 64 in the lovely country of New Zealand – Who knew we would go that far out to have Rachel implemented here. But again, Piers made that connection. He came up with a couple of names. We found a real synergy with Spark 64. And again, they were completely in sync with what our vision was with Rachel 1.0 and 2.0. And they worked at it and they got it live pretty quickly and with reasonable costs. So, yeah, huge shout-out to them as well. They have been an integral part of getting Rachel off the ground and we would love to, of course, continue working with them as we build the next versions of the platform.

Tyler: Great. So, Spark 64 for a huge thanks from us at UneeQ as well. That’s part of our partner ecosystem and anybody on this call looking for a partner, Spark 64 would be a great partner to work with. You know, Shashank’s given them the big two thumbs up. So with that, I just want to say thanks so much for your time. Looking forward to seeing Rachel 2.0 and where Arcus grows from here on, man.

Shashank: Thank you, Tyler; and thank you for making Arcus a part of this.

‍