The Use Case Podcast: Storytelling About Amplio with Trung Tran

Music:  00:02
Welcome to Recruiting Daily’s Use Case Podcast. A show dedicated to the storytelling that happens, or should happen when practitioners purchase technology. Each episode is designed to inspire new ways and ideas to make your business better as we speak with the brightest minds in recruitment at HR Tech. That’s what we do. Here’s your host, William Tincup.

William:  00:25
Ladies and gentlemen, this is William Tincup. You’re listening to the Use Case Podcast. Today, we have Trung Tran from Amplio, and we’re talking all about the use case, the business case that his prospects and customers make for purchasing Amplio. So, without any further ado, let’s get into some introductions. Trung, would you do us a favor and introduce both yourself and Amplio?

Trung:  00:48
Sure, I’d love to. I appreciate the opportunity here. My name’s Trung Tran. I have worked for Silicon Valley for 17 years. I’ve built 72 products with 2.2 billion in revenue. I’m really focused on technology and what it means how technology affects our lives. I got tagged to be a DARPA PM to work an AI program that encourage AI platforms for DARPA. And for me, AI’s a great opportunity because it’s the first time we have a technology to help us to think better. A lot of people think of AI as something to replace us or replace people, but really, as a way augment our thinking and giving us access to more information, I think AI can be a huge tool for what we want to do, that’s kind of leads to what Amplio is.

Trung:  01:34
Amplio is a company, is by helping people make better decisions, in particular, we want to see and help people interpret their data. A lot of companies that are in the data business, they say, “Oh, you don’t need your data, you’re too stupid to use it. We’ll tell you what to do.” Right? And the answer is no, that’s not really fair. It’s their data, and they own it, and you need to help them understand it so they can utilize it better. You don’t just take it from them. And that’s kind of our goal is to how to help people understand their data, interpret it, understand the reasoning behind interpretation, and actually use it to their benefit.

William:  02:13
I love that, and for folks that don’t know DARPA, I had it when I was in DC, interning for this Smithsonian, I had a friend at DARPA that interned there, and that was in the ’90s, and they were talking about stuff that we’re just now seeing now. So, for those who don’t know DARPA, go ahead and google it. It’s amazing because you’re working on technology and things that, again, people don’t see or commercialize until much, much, much later.

William:  02:43
So, let’s sink into Amplio in terms of the inside. I love the idea of you getting us to re-think AI as not just a replacement but giving us actual insight. And I think you touched on it, but just after we get the insight, then the recommendation in terms of what we should do with the data. Now that we see the data, and maybe we have some insight around the data, what are maybe even if it’s multiple paths, what should we then action? What should be do?

Trung:  03:20
Well, it’s obviously correct. I mean, I think, one of the problems AI and data analytics space everyone’s focused on gathering the data, and they are presenting or visualizing it to the user, and then assuming the user knows what to do about it. On the fitness side, one of the things that we think is funny is this whole idea of 10,000 steps, you know, from Fitbit? And if you research 10,000 steps, it was in the 1964 Olympics, where a company wanted to build a pedometer, Japan, and they called it 10,000 Step pedometer. And because they called it 10,000 Step pedometer, 10,000 became the health standard for [inaudible 00:03:56]-

William:  03:56
Oh, you just broke a bunch of people’s hearts right there. You just told them Santa Claus doesn’t exist.

Trung:  04:05
Right. So, that’s the thing. You should always view this data, and you don’t know if I walk 10,000 or 20,000 steps, do I feel better? What am I supposed to do, right?

William:  04:13
Right.

Trung:  04:13
How am I supposed to feel? And then there’s think of predictive analytics. They’ll predict if nothing changes, this is what’s going to happen, right? Nothing changes, just who’s going to election, right, kind of thing?

William:  04:23
Right.

Trung:  04:24
But the real question we want is if an asteroid’s coming at us, William, we want to know to do about that, right? We don’t want to just accept the fact that the asteroid’s going to come and kill us, we want to course of actions that maybe change that possible future and will help us get better, overcome our problems. And that’s really where AI becomes really interesting because having the ability to see many possible futures and help us decide the one that most likely will get us to our goal, and help us kind of make decisions to get there is kind of where I think AI really has the power, help us actually thinking a lot better because there are a lot of options. There are a lot of things that can confuse you, and most of the time, we wing it. We do trial and error. We jump from one solution to the other without a real sense of how likely doing that thing will get us where we want to go. And I think that’s sort of the beauty of AI, and that’s the AI that we’re building is to kind of help us simulate those futures and choose the best one.

William:  05:22
I loved the way you framed it up for folks is that there’s levels of this. So, yes, you need to collect data, and yes, you need to visualize the data, and yes, at the next level, it needs to give you some insight in some decision-making. Great. And you touched on it at the very, very last moment in terms of scenario planning. In a perfect world, it would then come back and say, given the data and given kind of everything that it knows already, here’s a couple of different ways that you can then action this.

Trung:  05:59
Yes.

William:  05:59
So, and again, if you’re sitting on a ton of data, it’s obviously learning, and the more data it’s sitting on, the more it’s learning, and all of that stuff given, stated, but do you see it more in the future used as not just a decision-making tool, but also, just more in the scenario planning?

Trung:  06:20
Yeah, I definitely think so. That’s kind of what our product is really focused on. When you look at the scenario planning, and you say it is incumbent that the scenario planning is done individually, right?

William:  06:32
Right.

Trung:  06:33
At an individual level because what a scenario plan is for the entire United States and an average person in the United States probably doesn’t apply to me, right?

William:  06:40
Right.

Trung:  06:41
So, big data analytics really doesn’t help there. Small data analytics are more important, and that’s what we focus on. We use wearable data, biometrics, psychometric data to kind of build up a digital twin of the person, and we say, okay, this is your digital twin. This is how you normally feel. This is how you normally feel when you’re doing well. This is what you normally feel when you’re not doing well. And these are the changes in numbers that seem to affect either one of those states, where you’re being successful or not, [inaudible 00:07:08]. And when you look at things like burnout that people are complaining about these days with the pandemic, and you look at things like anxiety, depression, what are the coping mechanisms to help you get my numbers back to where they were when I was happy and performing well?

Trung:  07:25
And so, for our sample, we build this digital twin, which is really a probability distribution that kind of tells you that based on all these factors, this is what likely is going to happen, and if you change these things, you do these routines, or you change these habits you have, you’re more likely to get to the state you want to be in where you’re happier and [inaudible 00:07:45]. And it makes [crosstalk 00:07:46]-

William:  07:45
Well, you’re getting to … I’m sorry.

Trung:  07:48
[inaudible 00:07:48].

William:  07:48
You’re getting to your most optimal, your most best version of yourself.

Trung:  07:54
Yeah. Also, it’s funny, we talk to a lot of programmers and other people who are passionate about what they do. These are an examples of a gamer, you can see gamers that play three days straight without sleep or something like that, or there are coders that do that or people that just are your high-performing type of employees that are just killing it kind of thing, but you do that enough, you get burnt out, and then you eventually don’t want to ever do it again. And so having something that reminds you, hey, maybe it’s time to take a break, William. You’ve been on your computer too long. Go walk around. Eat dinner. Having something that kind of reminds you and kind of gives you that kind of advice, and to give you those kind of triggers so that you can balance yourself a lot better is not something we’re taught. When I was going through school and all that, it was just work, work, work, work, right, kind of thing?

William:  08:41
Right.

Trung:  08:41
But you never talked about self-helping yourself or making sure you know when to back off. And a lot of that scenario planning is just how do I manage myself so I don’t get burnt out and I don’t get too stressed out?

William:  08:53
It’s really interesting because, and I want to get into the digital twin part in just a second, it’s really interesting because you’re taking in a lot of different data, and that can change moment to moment, day today, week to week and person to person, obviously, and so, what you needed and what you need on Tuesday depending on what you’re doing on a Wednesday, you might need something different.

Trung:  09:18
Yes.

William:  09:18
And again, that’s getting back to some of those data inputs and being able to analyze that. If you’ve been sitting for eight hours on Tuesday, you might need to get up and walk around, but let’s say on Wednesday, you’re at your standing laptop and you’re walking around. It’s like, I can see it being both proactive and reactive simultaneously to then understand kind of where you are and what you’re doing, and then making those recommendations based on real-time data.

Trung:  09:51
Yeah. And like you said, if you think about it, us older guys, you’re probably a lot younger than I am, William, right?

William:  09:57
I’m not touching that. Now see, you said almost 20 years in the Valley?

Trung:  10:02
Yeah, yeah.

William:  10:02
Those are like dog years. So, really, it’s like seven years. So, you’re at least 70 years old.

Trung:  10:08
Yeah, that’s true. That is true. But who you were at 20, and well, you can work out and you can work hard, you can pull all-nighters and work really hard in your 20s. In your 30s and 40s, now that’s harder to do.

William:  10:22
Right.

Trung:  10:22
Not only you have family pressures, but physically, you can’t do it.

William:  10:25
Right.

Trung:  10:26
And your emotional state in your 20s and your emotional state in your 30s are different, right? Well, we kind of think the digital twin is someone that is you, that evolves with you. Does that make sense, right?

William:  10:36
Oh, yeah. Okay.

Trung:  10:37
So, then you know this changing is your changing. And so, it’s not giving you recommendations like go run 15 miles when you’re 50, you know what I mean?

William:  10:45
Right.

Trung:  10:46
It may be something you can do in your 20s, but hey, this is more realistic, this is kind of what we know will make you better and not get you hurt or anything like that, and that’s the thought process for us is that there is so much data you can collect it from wearables from you, that we can build up this distribution and kind of learn how you are. Your digital twin can be you. We only want to learn about your data because anything else is confusing. I can’t compare you to me, or I can’t compare you to the next guy because we’re different genetically. We grew up in different environment, kind of thing, and so what I need and what you need is probably different things. So, we don’t pollute ourselves with anyone else’s data, we just use your data.

Trung:  11:25
And from a psychometric standpoint or a mood standpoint, John Hopkins did a study called Mood 24/7, and it’s actually now a product, where if you ask a person one question a day, are you happy? You’re actually able to track over time and predict suicide rates very accurately because every day, you may not be [inaudible 00:11:46] but every day as you’re answer the question, you start building up a trend and a dataset that says, okay, this is when he’s really happy, and this is when he’s really not happy, right? And you start trending and start saying, okay, then [inaudible 00:11:57] commit suicide.

Trung:  11:58
That’s kind of what we do. We take a biometric, and we do these daily psychometric surveys to understand your mood and your mental state. We use time [inaudible 00:12:07] see how focused you are and stuff like that, and we do that on a consistent basis, and we kind of profile you. We learn who you are, mentally, physically and emotionally. And then we start deviating from that, we can actually step in and start simulating all these [inaudible 00:12:21] factors that get you back on track, and that’s the key is that we have a ton of data on you that’s been collected, but it’s basically being dropped on the floor right now because we don’t know what to do with it, but if we build that model that [inaudible 00:12:35] with the digital twin, we can do a lot with it because we really want to know what works and doesn’t work for you. We want to know you. That’s what we’re learning. [inaudible 00:12:43] machine learning, and machine learning [inaudible 00:12:45] trends. While we’re learning the [inaudible 00:12:47] trends of what works with William and what doesn’t, and that’s how we’re going forward with our product.

William:  12:53
So, a couple of things real quick, and you’ll probably have to break this down in real simplistic terms for me and the audience, but you got data on the person, I’m assuming that you roll some of this data up in aggregate and anonymize it so that, and again, that case where you’re not going to recommend a 15-mile run for a 50-year-old in a 100-degree heat or something like that. But explain to us kind of how those recommendations, obviously they’re tailored to the individual because you’re gathering data from the individual and about the individual, and so you’re building that digital twin that kind of moves with them and changes with them, but I’m also assuming that you’re getting data from all of these other places, and then it also helps you feed recommendations in. Am I wrong about that? Or do I-

Trung:  13:50
You’re right. I mean, there’s two things. The first thing you’re talking about is bounding this solution set, right?

William:  13:55
Right.

Trung:  13:56
And so, we do need to know general population information like what do general 50-year-olds can do or not do, or what should be they HRV heart rate variation, the thing that measures your recovery, I think we call it your readiness score. A lot of people have different names for it, but what you’re HRV is when you’re older, and when you’re younger, and so we do have those population bounds, so we’re not doing something too crazy. But the thing with digitalization and the thing about what we do is we don’t learn you once, we continuously learn you, kind of thing, right?

William:  14:28
Right.

Trung:  14:29
So, if we give you a recommendation, William, and we see that it’s not having the impact we expect it to have, we ask you to go to sleep two hours early, you do and you don’t seem to perform any better, we’ll start modifying how we believe these interventions work for you, these general kind of things. And you may be a 50-year-old that can run 15 miles in a day and us asking you to run one or two miles doesn’t seem to move the needle, we’ll start switching up your routine or evolving it, so you’re starting to run up to the 13 miles or whatever seems to have an impact on what you want to do. Because that’s really what it is, I mean if in minute detail we can figure out whether you’re getting improvement or not, we can double-down on the things you’re improving on, right? Generally, like it’s you and me who do trial and error. It takes us multiple weeks, maybe even months to figure out if the thing we’re trying out actually works or not and a lot of times we give up because we don’t know if there’s an improvement and we give up just before it actually gives an improvement we want.

Trung:  15:24
And so, from our standpoint, we’re monitoring you and constantly looking at hey, we had you try this, it didn’t have as much effect as we’d like so we’re going to switch it up to something with more impact, then you’re constantly being tailored to like I said, what works and doesn’t work for you, and just constantly looking to see, I gave him a suggestion, did it work? And if it didn’t, we’re going to put that feedback in, adjust them out and give you a better recommendation. And over time, we’re just going to be spot on, right?

William:  15:50
I love that. I love it. And again, yeah, the more that you know in real-time from the data that you know about that person at that particular moment, the recommendations are just going to be spot on. So, there’s a couple of things, one is when people do a demo, when they first kind of see Amplio for the first time, what do they kind of dazzled by, what are they amazed by, the a-ha moment? Like what do they see and what kind of turns them on about the product?

Trung:  16:28
The first thing is they go, they don’t see graphs. We don’t give them any data or overwhelm. We give them say routines, we integrate into their calendar, and just try to stay. We work with a few of the Major League baseball players, and what we found out is that they said, “Don’t tell us the data, just tell us what to do.” And we live in a society with so many options. Going down the cereal aisle, I remember when I was younger, there was maybe one or two types of cereal, right? And now there’s thousands. And that kind of decision fatigue pretty much for our entire lives, and so when you have a bunch of data shown to you, and you have to interpret it and figure out what to do, that’s a lot more mental energy than say, tell us what to do and tell us how we’re doing? And are we making progress or not? We’ve always [inaudible 00:17:16] all customers would jump into and dive into data if they’re quantitative people, but the majority of people just want to know what to do, and we try and make it simple and clean. We just tell you, hey, these are the things that we think will make you better, and this is how much progress you’re making or not making on that, and these are some of the changes we’re making because of that.

Trung:  17:35
And also, we care a lot about … We make our recommendations based on three things. First, an onboarding session that kind of tells us kind of what your personality’s like and what you like and dislike. We’re not going to ask you to go eat kale if you hate kale, kind of thing. And then second thing is sometimes people vote by their feet, and so we look to see how compliant you are with the recommendations we gave you. There’s one reason why a recommendation won’t work because you don’t do it, right? And so, if you’re not doing it, then we will adjust our recommendations based upon, hey, let’s try something else because you’re just not doing it. We asked you to do it for two weeks, never did it once, we’re going to change it up for you. And three, whether it has an impact or not. And so, those three things are kind of what we build into the recommendation. That’s the experience you get is you get, hey, they’re already taking my input. And you see a lot of products like this that just tell you what to do. is more sleep, right? Well, I live a pretty busy life, you probably do too, William, getting two extra hours of sleep is like a bridge too far for me.

William:  18:40
Yeah. Yeah.

Trung:  18:42
I’d like to. I mean, there’s a dream.

William:  18:43
Yeah, who doesn’t?

Trung:  18:45
Yeah, but it’s not possible for me to do, and so a couple of weeks, you keep telling me to go to bed two hours early, you can tell me forever, it’s never going to happen, and so the question is, now that we have an AI can it learn that hey, Trung’s never going to go to bed two hours early, let’s give him some meditation, think of some other ways, maybe go get him to take a nap during the day to get the kind of rest he needs for his brain to kind of re-energize and flush out, and not rely on the gold standard answer of getting eight hours of sleep, you know what I mean? I don’t know how many adults get eight hours of sleep. Probably a pretty few. But the-

William:  19:22
It’s none of my friends. But yes.

Trung:  19:23
None of my friends, yeah. That’s the thing, I mean the idea that well, while I’m going to say, hey, you’re already taking my input. Now, in the beginning, the reason why I say we liked the onboarding thing and we kind of look to see what you do and don’t do, I said at the beginning what we see because we’re going to build up data to learn what works and what doesn’t work for you, and then you [crosstalk 00:19:44] better-

William:  19:43
You said at that point you said … Sorry Trung, are you using assessments or you take them through kind of some of the personality, behavioral, those types of assessments or do you all-

Trung:  19:55
Yeah, so it’s part of the onboarding we do. We do OCEANs, which is a pretty standard personality test. We do some cognitive testing, and again, it’s just to bound what … Like some people, like one of the baseball players had as super logical, but he didn’t think verbally, he did terrible in verbal exams, so again, understanding kind of what your limitations are, what you’re like is really important to us, so we wouldn’t force you to take verbal cognitive tests because that doesn’t tell me whether you’re focused or not, you just don’t like doing those tests, right?

William:  20:27
Right. Yeah, I can see understanding or having some type of, at least, basic understand of IQ and EQ, and learning styles and, like you said, behavioral and personality, a blend of all of those things. So, you start with some data so that you start to learn the person, so you can build that digital twin.

Trung:  20:46
Right. And that’s the thing like you said earlier, which we can bound it, a lot with like these tests in the population, things, the interpretation of the test results, we can start giving you recommendations right off the be able to that are tailored to you in a little more general sense, but they’re not just things that we pick out of the air, do you know what I mean? Like, get eight hours of sleep [inaudible 00:21:05]-

William:  21:05
Right

Trung:  21:06
Now, based upon what we see is your personality and who you are on the onboarding. And then, as you use the product more often, I always tell people that digital twin’s kind of like your pet. You’re in charge of its care and feeding. The more you do things, and you answer the surveys, the five-minute survey every day, and you wear the wearable, then the better the data will be, you know what I mean? The more robust it is.

William:  21:33
And the wearable can be a bunch of different types of wearables?

Trung:  21:36
Yeah, we support Garmin, Fitbit, Oura, SHOOP, and EMOTIV if you want to really do EEG and that kind of testing, but yeah, we do support. We don’t support Apple right now because Apple requires us to build an app because they don’t broadcast these in the cloud.

William:  21:56
That’s right. Yeah.

Trung:  21:57
We don’t quite have that skillset yet, so we have to hire someone to do it, but as soon as we get that done, we’ll be able to support Apple as well.

William:  22:03
Yeah, sure. So, two questions on the way out. One is what is Amplio connected to, like to other technology? So, once we’ve got the digital twin and you’re making people better, in a work environment, let’s say, do you-

Trung:  22:23
Yeah, that’s our primary customer. We start off as kind of a benefit from a wellness standpoint to give to employees to get them to be better. But the pandemic has really rewritten the rules between the employer and employee. Remote work. People want people to be more supportive, view them as a whole person, not just as a performance number kind of thing. And in a distributive environment, that’s super hard to do. This is where we aggregate data for the employer, and that makes it sound like Big Brother, but they don’t actually know who you are. They can’t ask and they don’t need to know. And the second thing is that the reality is that 80% of Americans rely on their employer for benefits.

William:  23:03
Right.

Trung:  23:03
Like when we work with Major League baseball players, they may have a cardiologist on speed dial, that there’s concierge doctor for them, right? Most of us don’t have that. We need the benefits from our employer, and we like the aggregated idea to the HR department to figure out how to provide better benefits. Anna [Doley], one of my friends works at Northrop Grumman, when her firm brought their people back to work last September during the pandemic, well, we’re still in the pandemic, when the pandemic first started-

William:  23:33
Yeah, yeah.

Trung:  23:36
… a lot of them were experiencing anxiety issues. My friend spent 15 minutes shaking in her car before she could go to work because there’s so much you certainty. And it took [inaudible 00:23:48] a while, but they brought in anxiety people, people that could kind of help with coping mechanisms and people could help if you have bigger problems than just anxiety. And so, giving the employer a little bit of insight into the group and what the group benefits would help the group the best is something we want to tie into because we’re not going to treat bipolar, we’re not going to give you medication anything. We’re going to try to keep you to cope. We’re going to hand you over the wellness programs, the AAP programs of your employer that can help you with additional things, and tie all of that together in a way that people can get the help you need. If they don’t need a lot of help, they stick with us, and if they need a little more serious help, they can get it as they need it.

William:  24:31
I love it. Okay, on the way out, favorite customer story so far without naming names, of course, but just your favorite story.

Trung:  24:41
My favorite story is we were working with a track and field team, [inaudible 00:24:45] track and field team and we were monitoring all these things to see if we could make their runners better, and actually, we improved their ranking quite a bit. But there was a girl on that team who was a highly recruited athlete, and she wasn’t doing well, and we were trying to figure out where her running [inaudible 00:25:01] were wrong, whether or not her sleep was okay, and was she spending time recovering. And it turned out that her nutrition had changed. Her testosterone got cut in half because, I guess, when she was growing up, she wasn’t allowed to drink a lot of caffeine or soda, and with the stress of going to school first time, she was just hammering it, and it does cut your testosterone in half. And it’s just as important as men because testosterone is the recovery hormone, it helps you build muscles and that stuff, and running is a pretty brutal activity-

William:  25:33
Turns out.

Trung:  25:34
… when it comes down to tearing down the muscle tissue and stuff like that, so she, in essence, was never recovering from all her runs. That’s why she was performing worse and worse. And though the coach at the time said, “I would’ve thought it was a kind of motivation issue and I would’ve made her run more.” And I said, “Yeah, if that happened, then the result would be either she gets hurt because she’s not rebuilding the muscle, or she’ll get frustrated and quit.” And just having the ability to say, “Hey, it’s not you. You’re trying as hard as you can. You don’t suck or whatever it is.” Right?

William:  26:02
Yeah.

Trung:  26:02
She stopped drinking the soda was a big deal, and it really changed her attitude and stuff like that because she was really worried that she wouldn’t be able to run anymore.

William:  26:08
It changed her life. I mean that type of insight, that was life-changing.

Trung:  26:16
Yeah.

William:  26:17
Because I’m sure, you’re right, it impacted her mental state, and even her confidence and stuff like that, so great story. Brother, you’re doing great work.

Trung:  26:29
Thank you, William.

William:  26:29
Thank you so much for carving out time for the Use Case Podcast.

Trung:  26:33
No problem. Thank you. All right.

William:  26:36
Absolutely. And thanks for everyone listening to the Use Case Podcast. Until next time.

Music:  26:38
You’ve been listening to Recruiting Daily’s Use Case Podcast. Be sure to subscribe on your favorite platform, and hit us up at recruitingdaily.com.