Found In The Rockies

Dr. Grin Lord & Brian Williams \\ Bringing Empathy to the World of AI

May 24, 2023 Les Craig Season 3 Episode 19
Found In The Rockies
Dr. Grin Lord & Brian Williams \\ Bringing Empathy to the World of AI
Show Notes Transcript

In today’s episode, we're gonna get empathetic. We are welcoming Dr. Grin Lord, CEO and founder of mpathic, and Brian Williams, who is the VP of Engineering and mpathic. Grin and Brian are going to talk about how they are working to bring empathy into the world of AI throught their unique API. 

Here’s a closer look at the episode:

  • Brian’s background and history with startups.
  • Grin’s early life growing up in the midwest and ending up in Seattle.
  • How Grin’s work in a trauma center inspired the idea for mpathic.
  • The magic door in psychology.
  • How chatbots can be trained to be more empathetic.
  • Brian joining mpathic.
  • Empathy rocks, the training game for therapists.
  • The science of empathy.
  • AI assisted fidelity monitoring to replace human doctors.
  • Understanding how empathy works in the sales cycle.
  • What excites Brian the most from the technology perspective.
  • How mpathic is different from ChatGPT
  • Looking forward to challenge ahead.
  • What is something in the future that is exciting
  • A vision for the future of empathy.


Mpathic website:

Grin LinkedIn:

Brain LinkedIn:

Mpathic LinkedIn:

Grin  00:00

Yeah, it's empathy is a pretty divisive word actually. And I didn't know that that founding this company, by the way that I understand empathy is accurate understanding. So, that's how we define it, if you accurately understand another person, they experienced trust and rapport and alliances and all these other things associated with empathy. So it's not like a warm fuzziness. It's, do you feel understood. And that's a relational understanding as well. It's not one person to sign in, I'm empathic and I did all the right things. So therefore, I have empathy, no, who that person needs to feel accurately understood in an alliance with you and building trust.

Les  00:47

This is Found in the Rockies, a podcast about the startup ecosystem in the Rocky Mountain region, featuring the founders, funders and contributors, and most importantly, the stories of what they're building. I'm Les Craig from Next Frontier Capital. And on today's episode, we're gonna get empathetic. I'm so excited to feature Grin Lord, CEO and founder of mpathic, and Brian Williams, who is the VP of Engineering at mpathic. Welcome to the show.

Grin  01:19

Thank you. Great to be here, Les.

Les  01:21

I'm so excited to have you both. And I'm actually today, I'm gonna start off with an apology. Is that okay? Can I do that? You didn’t see this coming. The apology I'm gonna make is I'm a bit of a cyber psychopathic host, which is the opposite of empathetic. All right, so that may be the most empathetic thing and behavior I exhibit all day. I'm teasing. I'm teasing. This is gonna be super fun. We're going to talk all about empathy. We're going to talk all about startups and where you're maybe you dabble a little bit in some AI, you think? A little bit

Grin  01:53

A little bit not too much, though. That is my trigger word. We're not going into GPT we're not gonna…

Les  02:00

What’s that? Let's talk about that, let’s go there. What is this GPT thing you speak of? All right? No, all right. That's not what this episode is about. I in fact, what this episode is about is stories. I want to hear your stories. Brian, you want to go first? Let's hear about you? Where you grew up? What brought you to this? This company? 

Brian  02:19

Yeah, so I grew up outside of Chicago, and I've gone through two startups and love working for startups. And both of those were successful next, and learned a ton of those, and one of the core beliefs, I kind of built at those startups is really leading with empathy. And there's been some great publications over the years, like Forbes and others on how important empathy is to entrepreneurship and got connected to Grin through a co-worker of mine that I worked at both both Elastic and Grafana with Katherine Johnson, and she is a co-founder and Grin can tell the story and how they crossed paths. But yeah, it got introduced to mpathic and Grin through her and kind of determine the opportunity to rejoin the startup world. Love what we’re doing and love working with the team here.

Les  02:58

 And so what about So you started in the Midwest? But you're not there anymore? Right

Brian  03:05

No, no. Yeah. Mount Denver. Yeah. Out on Denver. Yeah. So, the first startup I worked for when we exited a, we got acquired by a company that's headquartered here in Colorado. And I've always been a big skier my entire life, so super fun to, to get, to move out here with that acquisition back in 2006 it was. And yeah, I've been now here since. It's a great place to live. 

Les  03:26

Amazing. What's your favorite? What's your favorite mountain in the Colorado 

Brian  03:31

I'm biased. I'm biased. So we we're actually, we had a home and we're actually building. We sold it and up in Winter Park. So I'm a little biased. My, my daughter's on the ski team up there, so, that is our go-to mountain and we spend a lot of weekends up there. It's not a bad way to spend a weekend.

Les  03:45

No, I'll agree with you. I agree with you both both in season and out of ski season. Beautiful area. 

Brian  03:50

Oh yeah. All seasons. Except for mud season. Mud seasons not, not.

Les  03:53

But anyway, we won't dwell on that. Yeah, that's the only way we could get him on the podcast. He said mud season. All right, you got it. Brian. Grin what about you? What's, what's your story? Where'd you grow up and kinda how did it all evolve?

Grin  04:07

You're probably going to have to stop me because as a psychologist, like we spend a lot of time reflecting on how we all grew up and evolved. So like when I was born, literally. So I was raised in Minnesota in Minneapolis, and actually, my father, his grandfather, or his father, rather my grandfather, I never met him, but he actually was raised in Bozeman, and my dad was born in Billings, and my grandfather built the airport out there the original airport. Yeah. And then my grandfather met my grandmother and they moved to Hutchinson, Minnesota, which is where there was a large glue factory, and that is what became 3M. But back in the day, my parents are actually my dad's like in his late 80s. So this is like, much farther back. Yeah. So anyway, so that's how they settled into Minnesota. I was there, also Midwest with Brian. And then in high school, they, they're wildlife filmmakers, by the way. So I spent a long time in an RV in the summers traveling also through Montana, but through all the national parks, and then in high school, they decided they wanted to move into the RV full time to do filmmaking full time. So I actually went to boarding school, first in Switzerland, then in California, and then eventually went to college in Walla Walla, Washington. And then I moved to Spain, and then I was deported from Spain. And I was deported to Seattle. And that's how I've ended up in Seattle. 

Les  05:45

That's where you ended up in Seattle. Wow.

Grin  05:47

This is a long story. Yeah.

Les  05:51

And some of the work you did in Seattle, you were at the University of Washington for a little while, right?

Grin  05:56

Yeah. Actually, when I got deported to Seattle, it wasn't a formal deportation that we can have a different story about that.

Les  06:04

I was gonna say, maybe that's like a different pod, maybe a different podcast.

Grin  06:07

It was a paper error. I was doing the right things. I was trying to do the right things anyway. Very, yeah, whatever. So I moved to Seattle, kind of unintentionally, and started working for Harborview Medical Center, mainly because it was next to my apartment. And I was like, I want to volunteer here. And I want to learn about the emergency department, I want to figure out, you know, how hospitals work and started there. And that eventually kind of started the origin story poor mpathic, as I was recruited into a research study there, where we were looking at folks that were using empathic listening in the emergency department and ICU, and started to evaluate some of the, you know, outcomes of doing that research, especially because we are a level one trauma center, and we get a lot of accidents, a lot of drunk driving accidents, in particular, as well, and kind of used some of these techniques to create, you know, transformative outcomes for patients. So that is how, like the story, you know, this company in some way started was when I started at the hospital there.

Les  07:07

I see, you said something that I want to peel back. And it actually Brian alluded to it as well. But I want to peel this topic back a little bit. Empathic listening, what is this? What is this thing you speak of called empathic listening?

Grin  07:23

Yeah, it's empathy is a pretty divisive word, actually. And I didn't know that until founding this company, by the way that I understand empathy is accurate understanding. So, that's how we define it, if you accurately understand another person, they experienced trust, and rapport and alliances and all these other things associated with empathy. So it's not like a warm fuzziness. It's, do you feel understood? And that's a relational understanding as well. It's not one person to sign in, I'm empathic, and I did all the right things. So therefore, I have empathy. Like, no, that person needs to feel accurately understood in an alliance with you in building trust. So we actually, you know, have a ton of different behaviors and psychological constructs and emotions that we model at our company. But we started with empathy, because it's the most important thing. You can understand your customer, you can meet their needs, you can understand your patient or your colleagues. It's like foundational in all relationships. So that's why we started with empathy.

Les  08:24

Yeah, what a superpower. I mean, it really is. It's foundational for I mean, in life, I would say.

Grin  08:30

It is. Yeah. 100%. And it's been fun working with engineers like Brian and other people that, you know, we've been studying the science of empathy for a long time as psychologists and we know, kind of like the formula and what are the key ingredients to improving conversations, and then having this very cross disciplinary approach with machine learning, AI, engineers, and then bringing them into the fold of like, how this all works has been a great process. 

Les  08:57

Very cool. And Brian, I gotta say, you said the word empathy in your intro, that maybe the first time I ever heard an engineer speak that word. It's, and I'm teasing, I'm teasing. But it's not something that you know, 

Grin  09:11

They get a bad rap. They’re highly empathic. 

Les  09:16

I know, I know, they get a bad rap. But it's like, for you, this seems like it's been part of your career journey, the importance of understanding this internalizing it as an engineer. Is that right?

Brian  09:26

I mean, that's absolutely right. I mean, it's really, it's really interesting for me, and there's a lot of parallels that I think just in my background, a you know, the last startup is that is a group of people that we've worked together for 20 years, and that's pretty unusual. And a lot of these foundational skills and building relationships and having trust and rapport and all these things, I think are directly attributed to the success that we had and working together over those 20 some 20 plus years. Yeah, I mean, it directly relates and, you know, you started touching on that a little bit Les. It's really exciting to take, you know, the psychological backing and the models and everything we're doing at mpathic and apply these to all these different verticals in these unique ways using technologies. Super fun space to be in and very exciting that we can do. 

Les  10:09

Very, very cool. And so Grin at this point you have you've started to sort of test out some different some of these Different theories and practice working at this at this trauma center. What was sort of the evolution from that to eventually starting a company? What are some of the immediate steps? Yeah.

Grin  10:28

So the trial, the first trial that we did at Harbor View, when we did this kind of brief intervention with folks coming in to the emergency department where we, we accepted them for who they were, we didn't tell them how to live their life or what to do. And we just listened to them with empathy. The folks that got that intervention had major drops, and their drinking and as effects help for the next three years, just from a 15 minute conversation, it a 46% reduction and readmission to the hospital, which saved Harborview millions of dollars, because this was a population of folks that, you know, get into accidents frequently. And that's compared to treatment as usual, which is telling people, Hey, you did something bad and shame on you. And you need to get back into it, why'd you go off the wagon, whatever. So learning about this approach, you know, really helped people improve treatment outcomes, but also had like cost effectiveness for hospitals. So I was actually tasked with scaling this nationally, to all level one trauma centers as part of another research trial and like initiative, currently, we estimated saves about $2.6 billion to have folks doing these interventions in hospitals at scale. But one thing we learned with that is that it's actually really difficult to train these behaviors. We did two day workshops, PowerPoint presentations, things like that, and found that you can't learn empathy from a PowerPoint like surprised. So like, you know, and we're seeing this a lot too. And like DEIB initiatives, where people are getting coached to talk differently. And they it's like very hard to just be like, look at this slide. Now do things different, you need to have like a coach almost like learning a sport, need to have someone listen to exactly what you said and say, Hey, right there, see my how you did that? Like, this is this skill you're using, here's the new skill that you need to use, let's roleplay it, let's do it. So I ended up being part of this thing called a Clinical Trials Network. It was one of the largest phone coaching studies to date, where we would take recordings of doctors and therapists talking to their patients and say, these are all the good things or bad things you did, and psychologists would hand correct these, listen to this, and then call the doctor up and be like, All right, here's your feedback. And that’s what we had to do in early 2000s.

Les  12:36

Now, some have a little bit better bedside manner than others, like I've been there. We've all experienced it, right? 

Grin  12:42

Oh my gosh, so true. And what a dataset. I could get on a tangent about that, too. I think we have this phrase and the psychology field in the therapy field called the magic door. And the magic door is when you go into the therapy room, and you close the door, and it's confidential, that everything in there must be good. Because training programs for therapists right now very rarely have actual evaluation. So like a surgeon, you watch what they do, you're not gonna go let them into surgery, unless they have a lot of people that look for therapists is like this private area. So particularly in psychology, we were really interested in applying this technology, because we really, we don't believe in magic doors, like, what are you doing in there? And let's understand it, and let's improve you just like any other medical doctor. So that was actually our first product was we automated feedback to therapists 2009. That was the first speech signal processing pipeline. In at least in academia, I mean, like things like AWS transcribe, like all the tools we have now didn't exist. It was that,

Les  13:45

People forget. It seems like it wasn't that long ago, but it's technologically it was it was a different age.

Grin  13:53

Yeah, it took us about like six hours to process, you know, a 20 minute call and get instant, real time performance. But that was like what we could do with the processing speed and the tech. And like it was incredible that they built that, you know, in house, this was a collaboration with the University of Washington, USC, UC Irvine, University of Utah, also that commercialized into our first startup where we were like, okay, we can do this at scale, and train people and give them feedback. And that, you know, is a largely academic project with researchers. But for me, I was like, I see the potential for this to help everyone like anyone can listen with empathy.

Les  14:31

Was that was that project that it transitioned to was that Empathy Rocks is that

Grin  14:36

No, so Empathy Rocks came after so well, yeah, I got so Lyssn was my first startup. And then left Lyssn went to Youper, which is a conversational therapy chatbot did empathy improvements to their chatbot as well. And there's some interesting research studies coming out of that right now showing how people trust these chatbots equal or more than humans because they feel like they can talk about things. So it's all another tangent. 

Les  15:03

I was gonna say that's like, that's a really that's a really interesting topic in terms of like, when I think of chatbots, I think of the opposite of it's like, not only not empathetic, but annoying. But if you can train it, is that what we're doing where you could train a chatbot essentially to be empathetic.

Grin  15:18

Yeah, we did at Youper. And and, yeah, I can't we did a lot of internal studies that showed amazing business outcomes when that happened in terms of people establishing initial trust and converting to the product quickly when they felt that initial report established. And yeah, they ended up publishing some research studies coming out of that, that, you know, not only was the bot effective at impacting health outcomes, like reducing anxiety and depression, but also that people felt that they could be more honest to the bot because it wouldn't judge them. And so they would talk about, you know, traumas or things or feelings and thoughts that they thought were not acceptable for humans to hear. So it was kind of an interesting thing, because there's so much pushback right now about this kind of AI and mental health and, and that also was that was 2019. That that was, I was part of that work. So it's a topical thing now, but it's been ongoing.

Les  16:11

I can imagine a chatbot like, Oh, it doesn't pass the Turing test, because I can tell it's a chatbot is too empathetic, a human being could never be that empathetic.

Grin  16:19

I mean, we're in that mode right now. So with LLMs, like, I can train an LLM on transcripts of the most empathic therapists that are recorded, and every one of their treatment manuals and every YouTube video they've ever done in training and in seconds, and have them be that in the conversation, so like we are, this is not sci-fi anymore, we can take the top quality. I mean, you could even extend this to like, you know, putting like, I don't know figures historically that everyone is like, Wow, I love this and putting them into one kind of bot. So you're right, at that, making it more human would be almost the opposite. Like, trying to make it fallible, and like, have an ego and like, yeah, it's like, offended by you, and it doesn't want to talk and it's like, you know, like, these are real things. So

Brian  17:12

Generating podcasts, that’s going to be the next one Les.

Les  17:15

Brian, you're gonna put me out of work here. Don't dare say that.

Grin  17:20

It’s happening, it's happening. So I'm gonna train a model on your podcast Les.

Les  17:26

Oh, my goodness. The world is not ready for that. That madness. Yeah. Brian, so where were you in all this in terms of like, when did you finally you said, you gotta you got an introduction? When Yeah, you're gonna get just lit on fire about the opportunity here and decide, oh,

Brian  17:45

I mean it was last year, I mean, yeah. So last year, I started to you know, contemplate what's next and started evaluate, like startups who in my network was, you know, had opportunities available and can make introductions, I said, for found half the time and ready to get back to, you know, small companies startup where I can make a big impact. You know, leverage, mostly what I've been doing for the past, you know, 25 years. So, yeah, I joined mpathic August, last year.

Grin  18:10

What Brian's not telling you, that's kind of funny, and is that we also interviewed his wife who applied for sales. She came before Brian, and was very excited about the idea of empathy. And I didn't actually know that they were together, I thought it was valuating them and like, I want to hire these people. And then at the end, there were like, you can only have one of us and that I didn't even know. 

Brian  18:41

Which would not have been the first time we'd work together, we should worked together Elastic.

Les  18:46

Oh, wow. Amazing. That's fun. So Brian, what gets you the most excited about, you know, somebody that you know, clearly cares about this category, this opportunity. But as a technologist, what do you what gets you most excited about working on in this on a daily basis,

Brian  19:06

Just the opportunity we have, right? I mean, there's so many different verticals, so many different applications of what we're doing. I mean, when you think of this, everything from you know, sales to medical, which we talked about today, here, getting into, like, you know, customer service, the list just goes on and on, where I think this woman's a market, the I feel, not a lot of folks that have spent a lot of time exploring and the number of different ways we can take what we have here at mpathic and really have an impact and just on how humans communicate, which is everywhere. When you think about that, that's just every everywhere and everything we do every day. So

Les  19:40

Yeah, I think about it to like in a world of like, where it's increasingly important to build trust we used it used to be a part of our daily lives, because we were, we built trust with the people like, like, you know, who are the people in your neighborhood, like, those are the people we interacted with. Those are the people that we build trust with in a very human connected way, personal way.

Brian  20:03

Physical presence. Think about how COVID has changes in the virtual world that we all live in now. And you know, just how that has changed things and had an impact and how we establish some of these, you know, things like trust and empathy and things like that. It's more difficult and more challenging in this virtual world that we are post pandemic.

Les  20:19

Yeah, I mean, the opportunity is big. When did Grin, when did it sort of develop then from sort of like this? It seems like directionally it's moving towards mpathic that, by the way, for our listeners, that's the name of the company. I keep saying mpathic, I'm like, if I don't spell it out without the E, they might not even know what we're talking about. Grin, why don't you tell us a little bit about that the name and

Grin  20:43

yes, mpathic with an And, yeah, so I guess getting back into that story, I kind of, I wanted to teach everyone how to listen with empathy, I wanted to leave kind of this academic world of hyper focus on people that already were highly trained and getting into folks that are not highly trained, like, yeah, customer service agents that are following a script or your boss, that's very, you know, non-empathic. So that's how I wanted to go. But in order to make that leap, I needed to build basically all new models and training data, even though I had it all in the academic setting, you can't bring that over into a commercial operation. So I created Empathy Rocks, which was the training game for therapists. And what we did was actually teach them on this game, how to respond in real time to different comments that folks were leaving in support forums on Reddit, or they had put them publicly out, you know, for people to respond to, we put those responses into the game and had actual licensed therapists respond to them, using empathy skills, like reflective listening, open ended questions, affirmations, and then rank each other’s statements and correct them. And in doing that, they would earn continuing education credits, and we would have at the this data flywheel of labeled data where no patient information was involved and that therapists were earning, you know their continuing education credits. So we were able to have hundreds of 1000s of data points labeled without necessarily having to source actual, you know, therapy conversations, or medical conversations or employee, you know, which eventually now we do have, because we're integrated. But that was how we avoided this cold start problem, and really created a dataset that could be used for this kind of grammerly for empathy idea of auto correcting people in real time. And that actually was one small feature on the Empathy Rocks game was like, it was like, correct me, like change, change this so that we could have better data. And I remember a moment when one of our advisors in our accelerator was like, is, is that is that the? Like, are you gonna do that? And I was like, yeah, we're just gonna, so yeah, that was kind of how it started. And then we chose to do this API first approach of giving other SaaS companies the ability to do real time correction and analytics of these many different behaviors, including empathy, but we actually have 90 different things that we detect in terms of behavioral markers of how conversations are going. And that's how it started. And yes, and then Brian, came last year, when I realized that, yes, there's definitely an ML and NLP component to this, like, this is an AI company. But actually, when AI companies are successful, they have some verticalization. And they have built a product or platform around themselves that, fundamentally is a software fee. And so I needed top tier, you know, engineers, and people that have been doing this and startup spaces to take kind of my domain expertise and like, psychological stuff, and really productize it into something that people would use would be enterprise ready, they have all this security, all these things that are really important when you're analyzing, you know, sensitive conversations.

Les  23:55

Yeah, for sure. So I like the concept, a really unique for listeners just to rewind a little bit grammerly for empathy. Pretty cool conceptually, to think about that. And also, you know, yeah, when I think when I consider, you mentioned, these sort of 90 attributes, or, you know, sort of some of the tagging that you do, you know, I think for the common person like myself, I It's like empathy. You know, what, when you see it kind of thing, like, oh, that's empathetic. But there's that the science behind this is much deeper than than just that. Is that right? Can you talk a little bit the science?

Grin  24:31

Oh, absolutely. And so part of the reason we started with empathy is because there is a science base to it. We know exactly what the words, phrases and behaviors are that lead to positive outcomes, even across cultures, like there's pretty robust literature around this. So one of the I'll first start not with the behaviors, but some of the indicators that a conversation is going well and is empathic, and one of the ones that's pretty robust is this concept of synchrony. If you really like someone and you're getting along with them, you'll start to synchronize and the words you use your pacing, even your placement of things like prepositions and adverbs, and like very unconscious things that you can't, you can't actually pretend to do this. Like you can try to use the same words, but if you really like them, unconsciously, you start to mirror them and get into this way. And we even notice that people that are like in power in the conversation will typically be the ones that synchronize To the person not in power. So like, it'd be like the manager to the employee or the doctor to the patient. It's not like a mutual thing, it's when that person's being empathic, they start to synchronize. So we measure that that's one of the things that we've looked at in behaviors that you really can't train. It's kind of either you have it or you don't. And we think of that as like, 10% of the slice of empathy, and that 90% of the other behaviors are trainable. And those are the things that we like to correct and prompt. 

Les  25:56

Oh I see, so if you really liked them, you start to synchronize, you start to revere them, use use. Oh, am I doing that right now? Oh, that's

Grin  26:10

I was like, is this a setup? That's gonna happen. That actually happens on our sales calls too. It's funny,

Les  26:18

On a serious note, I mean, it's Yeah, makes sense. Like we all can, I think we experienced that. And so for the other 90%, though, that's the 10% sliver for the other 90%, which is, I mean, a universe of things. How do you pick apart that problem? And maybe Brian chime in technically about how?

Brian  26:37

Absolutely. I mean, from a technical perspective, it's models. And I think one of the most exciting things that we have is the fact that we have a, you know, top tier expert labeling team that are trained clinically and rooted in psychology, looking at data from perspective. And I think it's one of our really unique value propositions is taking that data and using that to train our models, build our models, and actually using that data in real world settings to do our detections through our API and to get that unique perspective on being able to analyze these conversations. And it's really exciting. And it's, you know, something unique that we're able to do at mpathic. And one of the reasons why I joined the company.

Grin  27:15

Yeah, the way that the API is structured that Brian is actually had a huge impact on is this idea of a modular detections of things that are going, you know, good or bad in the conversation. So we detect the different behaviors. Examples of that would be like, detect, when you're making a confrontation, I detect when you've given me an affirmation or supportive statement, you know, we're detecting all of those things, then we give them a tip, which is typically vertical specific, and expert led. So for sales that would be like to close the deal, do this, or you know, a medical setting it by be, try using an if statement to like, express your concern. And then we give a correction, which is this generative AI piece where we're taking what they said, and say, Let's translate that confront into an I statement. And here's how you should say, what you intend to say. And like you can accept or reject that or use that in the conversation next, use that as a highlight, like all of our customers used the tips differently. But this kind of three part structure, Brian's been really instrumental in kind of creating that so that our customers can decide, do we want to just detect problems and good things? Or do we really want this real time coaching and like go the next step all the way? 

Les  28:27

That's what I wanted to emphasize. We're talking real, this is real time you could

Brian  28:32

You could be doing it in this podcast Les, like you could be getting the real time feedback. And like, are you establishing synchrony, like all these metrics, all these 90+ behaviors?

Grin  28:40

Actually, we should rewrite we can do that. Brian, let's run this afterwards and produce a report.

Brian  28:45

Yeah, we literally can. 

Grin  28:49

Alright, nevermind,

Les  28:52

Oh my goodness. We'll put it in the show notes. I guess. It'll be the cancellation of season three. But let's do that. I'd love it. I think it'd be really cool. I think I'd be really

Grin  29:00

We have. We have some models, too, that we're developing that I'm noticing are happening on this call. So not just agreement, we have this thing called synchrony moments. And it's where people when they are like really getting along, they start to repeat each other a little bit. So it's like another form of mirroring. So you'd be like, yeah, yeah, I got it. Right. Right. Right. Right. Yeah. Like, I'd like things start happening where people say little phrases and like, get excited….We would have a lot.

Les  29:28

I knew I was going to get psychoanalyzed on this call. Yeah.

Brian  29:31

The next podcast will be about about engineers working with a team of psychologists.

Les  29:36

That would be fun. So what are some of the industries I want to I just like, is there anything you could I mean, the, philosophically such a cool company, and I'm not just saying that because we're investors, but I mean, it is such a cool company. And so many applications, where are you finding the most traction? Like, you know, with some of these early use cases, like where's the impact high and the, you know, the conversations with customers, you know, exciting? 

Grin  30:04

Yeah, so given that I cut my teeth in the medical space, we did do some medical applications. One of the first things we did was aI assisted fidelity monitoring to replace actual humans and doctors that are evaluating conversations to say, is this medical conversation going well, or not? So our AI has been deployed with a customer at scale to do that, where we've shown a 10x cost savings compared to actual human doctors reviewing these calls, and we can review 100% of sessions and like, for example, a medical trial to assist with some of that and flag problematic areas where it's typically we'd only get 10% of those calls reviewed by humans. 

Les  30:47

It's cost prohibitive, right?

Grin  30:47

It's completely cost prohibitive, which is why folks like the FDA and others don't require that they say you're gonna have 10% spot checked. But with AI, there's this possibility. And just to be clear the FDA does not allow for AI to do this on its own, it has to be assisted right now. But there's a world that's better for both the consumers and companies that are doing, for example, drug testing, and things like this, where 100% of your trial can be evaluated at scale, in real time. So that's one application that, you know, has been evolving for us. And we're like wanting to explore more. But initially, we validated a bunch of markets, we looked at HR SaaS platforms, we have one company humanly here in Seattle, they use us to power their recruiting platform, they've had an 8% increase in candidate acceptance rate for folks that are more empathic and that are taking these empathic suggestions. So we looked at that, and then now we finally come down to sales, and are starting to predict the close rate of sales and use real time coaching at scale. And that's in the next 24 months, what we want to start to dig into more is understanding how empathy can work in the sales cycle, especially because with the advent of GPT4, for a lot of I hate to say this, but a lot of sales reps are going to be replaced, like, we're probably going to have a lot of email campaigns that will no longer be handwritten. And so the part that will matter in the sales cycle is that empathic connection. So one of our customers that we're deployed with, I guess, kind of a blend between sales and customer service is actually in insurance claims. And when people call in with like a car accident, or they've lost someone, and the first person they have to talk to as an agent, on the other end, you know, there's only so much that you can do by saying, Well, we're going to reimburse the car, you know, that person has to express immediate empathy and understanding just like they would in a medical setting. So the commonality for a lot of our customers is this sense of really enhancing humans to do what humans do best. Rather than automating them away. We know that that's coming in, and it's going to happen in text and things. But in the conversation, the conversation is the last frontier, right? Like, no one wants to lose humans, but humans are very variable in expressing empathy and some of them don't do it that well. So like we're trying to help them do that better across these kinds of use cases. It's a horizontal product, but we're starting with the sales medical proving that out and we have some great HR SaaS as well. 

Les  33:13

Yeah, it's great, great, great use cases. Very exciting. Brian, what are out from the technology perspective? What excites you the most about just current trends? And you know, kind of enhancing, I would assume, where you're already going and what the opportunity is there to just capitalize on on the

Brian  33:33

It's such an exciting time, I think, I think you said earlier last year, it seemed like 2019, or something like that was was a long time ago in this space, it seems like last week was like, Well, what's happening, you know, we have folks like open AI and others, they're just really kind of, you know, paving the path for these conversations around how AI can be used in different ways is great for us, because they're, you know, starting this conversation, so that's probably most excited to be I'm excited to see what happens in the space rebellious and how that gets commoditized. And, you know, like, there's, we've seen this before in technology and a lot of different areas. And it's going to happen again, in the AI ML space of like, who is going to win this battle of commoditizing, large language models, and really make it available for folks like mpathic and other enterprises to, you know, be able to let and do really creative things with and have it kind of be the next thing that we see all this innovation happen on top of and do great things with. So that's the most exciting to me, and it's changing weekly, if not daily,

Grin  34:31

I think a lot of startups are in a similar position to us where we'll just step back a bit. We were early partners with Open AI in 2020. So we had generative language capability, which was some of the magic that when people would use mpathic, they were like, Oh, if we what, how did it do that? And you know, and now, you know, a lot of people understand the concept of rewriting something with generative AI

Les  34:54

Less than explaining to do for sure.

Grin  34:57

Yeah, 100% and but now there's like a lot of competitors to Open AI and there's like new applications. There's auto ml self prompting ml, it's evolving. So a lot of startups right now, I think in the AI space are trying to figure out where to go next with their products. And we have a head start because we've been working on this for a while with LLMs and like I've been doing this but I think many of us are in a stage right now with our products of like, okay, how do we make this turn key specific to a vertical so it's accessible to people. But as soon as we make a decision on like, alright, here's the LLM. And here's the play. And here's the AI. It's like prompt commits like nothing. And like we made this joke this morning, I'm like, Okay, on the senior leadership team only, like one hour spent on AI updates in the news. Otherwise, we won’t build the product because there'll be like, Oh, no, here's this. And here's this. And so we're just kind of like in this mode of continuing to do what we do best, which is that the domain expertise, but like, these tools are, yes, super fast. So folks like Brian are just in a constant, flex of like some, here's our core product, and like, how are we going to now integrate the best in class? 

Les  36:13

Yeah. What about so I think about when I introduced my mom, I showed her chat GPT and she didn't really understand it. But then she finally got it. Like, I think if she, by the way, hi, mom. I know she's listened to the this episode. Like if she listened and she listened to this episode, she would probably be like, Well, wait a minute, like, mpathic does this, like I could go in chat GPT and say, I'm trying to fire an employee. Here's what I'm going to tell them. Make it empathetic. And it'll do it. It'll do it. But you know, so my mom asked me a question for a mom, how's this different? 

Grin  36:49

There's multiple ways it's different. One is like GPT is trained really well on documents. So like, if I were to write an email, or really like a job description, or something like that, like, it's very good. Conversationally, and I don't know, if you've ever listened to like a David Mamet play, or like some sort of conversations like, it's not the input and the output, they look a little bit different. And the real time nature of like how the suggestions and corrections are given, it needs to be quite fast. So what happens with just like doing a large prompt like that is it's fairly slow. The training data comes from documents. And one of the other things we see with GPT4, or when it doesn't use, by the way, we use GPT4, for in our product. So we build on top of this, but we have layers, what makes our product different is we have these proprietary detections that are built by psychologists like find this behavior, find this exact thing that is occurring in a medical conversation, it's definitely not part of the training data of GPT4, nor is the domain expertise around like this particular way that for example, like psychosis has finally started the show with that, but we actually have models of things like unusual reactions looks like these are not things that GPT4 understands how to detect or like handle in conversations. But we can leverage it by feeding it those things and saying, Okay, in this conversation, we would like to take this detection and like translate it into this other thing and have GPT see the examples from us and generate it at a sentence level, when you just put in a whole document, like what you're talking about with like, I want to talk to my employer, it will do something and it's likely going to start to be repetitive. So if you had, I'm just like, this isn’t an actual situation, if you had a manager who was like, we're going to send out performance reviews, and everyone was like, GPT, right, my performance review for me that's empathic, there's going to be a lot of repetition, in in that over time, and it also will optimize for being overly polite, which is isn't necessarily empathic, like has a slightly different definition. So anyway, long story short, we've optimized it for conversation, novelty and specificity for the exact behaviors, which is very hard to do with one generic prompt. And it's a little bit harder. But I mean, I'm, like, just being very humble here, like, this technology is evolving extremely fast. And so one of the things that we do is we're constantly evaluating our proprietary models against these large language models to see like, did it learn? Did it catch up? Where are we differentiated? How do we use it? Because there's certain things we detect like an open ended and closed ended question that, quite frankly, it's not a super big you know differentiator between something like GPT, but our more nuanced models are, and the amalgamation and the interpretation of that into a product is ultimately what we're offering. But I love that your mom is doing that. And she definitely should. And I encouraged everyone, like, use this product, like see what's possible now. Because now now you have the ability to extend yourself and your abilities and your brainstorming with like another, you know, model to help you.

Les  39:56

You know, my grandmother was still on a rotary phone when I introduced her to the cell phone. And so like, I'm just trying to keep my generations of people I love, like up to date, you know, but anyway, what I got a question for both of you, that I want you both to take a shot at if you're up for it, and that is what has been sort of an unexpected challenge along this journey of building a company like this. And, and second, actually, so it's two questions for each of you. And second of that is what is something that in the future that you know you're really excited about that may maybe also be a challenge, but something that you're excited about pushing forward into over the next you know, year or more, Who wants to go first. Right?

Grin  40:38


Brian  40:42

Let’s see if I remember your questions.

Grin  40:45

It was stacked. 

Les  40:47

I'm trying to give you a little time. Well, just take the challenge an unexpected challenge. 

Brian  40:55

Sure, I’ll take the challenge. I think the biggest challenge is, you know, API first enterprise, SAS sales are challenging, right? Like we're going after, you know, large companies and selling an API. That takes time. And that certainly is a challenge that pays big dividends in the end. And I struggle a little bit with it, because it's also challenge. The last startup I did was also a b2b enterprise API first SaaS company. So it's a challenge I'm personally familiar with. And that's probably why I hesitated a little bit on answering with that. But I do think it's one of our challenges. But it's a challenge that I think is exciting. Like, it's not the dividends make it absolutely worth it. In the end. It's no just things that we need to tackle in the right way with the right approach. And we'll be terribly successful in the end. I don’t have any doubts about that. Now, you gotta remind me of your other question.

Les  41:41

Well, hold on that I will go to Grin for her challenge. But yeah, that's great. 

Grin  41:46

Yeah. For me, as a psychologist, having never been a venture backed founder, I had a unique challenge of having to learn business really quickly. Psychologists are great at that, by the way, I mean, most of us are small business owners because they had like a private practice, or we worked in medical billing, like, by date is a completely different ballgame when you're dealing with venture capitalists and C corps. And the level of learning is, you have to be extremely fast at learning that stuff. So for me even just like abbreviations, were a challenge, understanding, you know, all of these words, combined with the fact I then also had to learn, you know, software engineering abbreviations, as well. And, and then I already have my like, medical ones. So just like three different languages.

Les  42:35

You got a whole alphabet soup? Yeah.

Grin  42:38

Yeah. So I felt like I had to take my medical knowledge, academia, turn it into business. And I definitely had some like, sputtery, it starts with like, I remember kind of trying to be like a business person and be empathy will increase your revenue and like ROI, and people like, Oh, gross, like, No, you can't save that, like, we don't like those words together. So it's like, Okay, empathy builds trust and loyalty and reduces you. And generally trying to find the business language of like, how to explain something that you already know, and saw, like a million times in your medical field, like, just like people like this, this matters. And then like, finding the business get behind it. So there obviously is one that you know, is there, but for me, it was more about like learning how to say, speak about this different world. 

Les  43:29

Very cool. That's great. Anything worth sharing about the future that you're excited about? It could be a challenge. It could be just something coming up.

Brian  43:36

I was gonna look Grin go, go for it Grin. I went first on the last one. 

Grin  43:38

Oh, yeah. I mean, for us, I am extremely excited about our applications. And these two verticals and sales, where we're seeing this faster sales cycle, real excitement about our ability to fulfill a promise that a lot of companies like not to call them by name, but things like you know, gone and chorus. They were supposed to help people in real time. But guess what, like, talk time topics that you said and was your tone good and bad is completely not actionable? Like, in an insurance call, it's gonna be sentiment is red and bad? Because we're talking about an accident. Does that mean that agent was bad? No. So like, we're trying to step in and add this psychological layer, this behavior shaping layer that actually fulfills the promise of replacing the manager and supervisor. And like, that is the part of this the most exciting to me to see how we can take this which we know has proven ROI and applications and medical and bring it into these other verticals and start to see that and we're already starting to see that. And the other cool thing that I'm looking forward to is, as Brian mentioned, kind of going on this API first journey, and seeing, you know, some of the go to market challenges that we're having, like, do we make a platform some of our customers are saying, Hey, I'm adding you on top of this thing that I don't like what why can you be that Just like I want you to be the platform. So so there will need to be some decisions that we make as the product evolves. But for me, I'm a researcher, I'm very data driven. I want to see people play with the API's understand how they're implementing, look at real time coaching and how it's working. And learn from that. But before we like, settle on these things, but just yeah, a lot of exciting things to come.

Les  45:23

Very bold, very bold vision for the future, but not surprising. Considering you, you are a finalist for Startup CEO of the Year. Oh, my Geek Wire award, just to say it just to say it, yes. Which, well, you're a finalist today. But by the time this episode airs

Grin  45:39

Well, now, either I'm not or, but you know, what's funny is a few of our customers are also in a nominated as CEO. So it's been funny, one of our customers pre he was posting on LinkedIn. He's like, I'm nominated, but vote for Grin. And I had, I had already put my vote in for myself. And I was like, perhaps, since I'm like, running to my husband, like, you're gonna go vote for free. And right now, it's because I didn't ever like tried to do this. But it's been great to have like, you know, a group of people that we're working closely with all nominated, and I'm looking forward to

Les  46:12

that. It sounds like you'll be in good company at the at the tea. Yes. It's lonely, lonely at the top. All right. What About Brian, anything you're looking forward to challenge ahead?

Brian  46:22

I mean, clearly, I'm looking forward to hearing the outcome of the award.

Les  46:27

Good answer. 

Grin  46:30

If he doesn't work for the CEO of the Year, it's gonna be a problem.

Brian  46:35

No, I would just echo what Grin said, The exciting thing about being API first is you get the API out there. And, you know, I've seen this play out before you get to see help Different people use different ways. And that becomes a driver for our company. Yes. Built in product and what mpathic becomes. So, you know, I'm just looking forward to getting what we can do out into, you know, folks here and seeing what they do with it. And, you know, being able to shape our future. All the creative things people do with what we have.

Les  47:02

I can't wait to see it as well. Brian, I mean, that's what I love about humans. If there's one thing we're gonna be creative, we're gonna do crazy things. So last question, I always love to kind of end on something more more personal and just kind of less, less businessy. But what either a story or a vision, a story of how empathy changed somebody's life that you want to share? Or, or just a vision for what it could do what it could do for us or for our world.

Grin  47:31

Yeah, I mean, I have a lot of many of the stories that I have, I can't actually share because I, you know, as a psychologist, you know, under research, but I can say that these skills are, are transformative. If you can learn, I often say this to people, when they're like, Well, where do I start how to become empathic? Like, it just seems like an amorphous thing. I often tell people, like just start with repeating back what you hear the simple reflection, you think you're gonna sound like a parent and a broken record. And like, watch what happens, go to a party, and don't ask a single question of anyone, just listen to what they say, and repeat it back and watch how they start talking. And like, watch what comes out. And, and as they start listening. So like these are, these are basic skills, they're just hard to do if you're not somewhat rigid about your application and stringency and to not talking about yourself and to listening and to like prioritizing this other person. And I've yet seen incredible transformation in families and relationships, you know, in, in, in my own life, for example, with, you know, folks that are in the engineering or software space, like teaching them these skills, and then watching them go into a meeting completely differently than that just go in. But I like the big vision, you know, for me, it's this, like, this core concept of empathy is like, could you imagine someone else as like? You or, could you imagine their children as your children? Like, could you actually do that thought exercise and that perspective taking and not to say that you will feel for them, or you need to be them or you're, like, a mind meld or something. But it's just this idea of, if we could all really accurately understand each other. And we could do that perspective, teaching how many problems in the world, you know, not exist. And there are a group of psychologists that do this, that do these kinds of radical listening groups and come in and bring in people that just have very different viewpoints and teach listening skills, and the outcome of those is consistently a love for the other person and this humanity and like, we know what to do, like, we know how to listen to people, but it's just a matter of getting some of these very simple, you know, training tips out there and integrating them where you are. So you don't have to go to a special training like that anyone can do this. So that's kind of my big, you know, change the world impact vision. 

Les  50:01

Beautiful. Thank you for sharing that. Brian?

Grin  50:03

Brian try and follow that now. 

Les  50:08

you already got you already got called out, you already got called out for taking your empathy into these engineering meetings.

Brian  50:17

I’m not a psychologist, so I can speak more freely. You know, for me is a little more lighthearted approach. I guess my 10 year old daughter came home the other day and she was really upset because her and her brother got into an argument because she felt like she could is about about, she felt like she could feel what the rock feels. And she was like really upset because he was like, it's just a rock. I don't know where you're talking about. You're crazy for thinking that you can reflect and understand and have these feelings about a rock. And she was just really shaken up about this. And I joke she's probably most, you know, the most empathetic 10 year old on the face of the planet. For me. I don't try to do something in my time here to try to make the world a little more empathetic. You're welcome, dude. Heart is

Les  51:02

so sweet. Thank you for sharing. Amazing. Well, listen, I just want to thank you both. This has been just such a such an insightful and thoughtful episode. So great to have you both on the podcast. And just to sort of wrap up Grin, why don't you tell our listeners a little bit more about where they can find you and Brian and mmpathic online.

Grin  51:21

Yeah, you can go to our website,  And that's the best place to go. We're also on LinkedIn there. We don't have a Twitter account or anything. So go to our website. And then you can also reach out to us at If you have specific questions or get in touch with either Brian or I.

Les  51:56

Thanks. Thank you both. Thank you for listening to this week's episode of found in the Rockies. You can find links in the show notes or go to next frontier to get transcripts, links, and contact information for today's guests. If you liked what you heard and want more, please don't forget to rate review and subscribe to get notified as our new episodes drop every two weeks. We'll see you next time.