Building What Couldn’t Be Built Before With AI

0:00Where AI and an AI coach can also play is that they can see your contacts. so they can see what's going on for you. A human coach can't sit with you at work and see what's going on. Welcome to the Talking AI Podcast, where we talk AI with both experts in the field and early adopters. I'm your host, Matt Page, and we're here to demystify AI for you so you can get some value from it. Let's talk some AI. What if the power of AI could fundamentally change how we grow, learn, and work, not by replacing human connection, but by scaling it? And our guest Today, Jamie Albers, co-founder of Mento, is reimagining the future of coaching and mentorship at the intersection of AI and human development.

0:39And Jamie and her team are tackling one of the big challenges out there in two-sided marketplace platforms and tools. It's that supply and demand mismatch. But instead of adding more human coaches, they're looking into the possibility of using AI to scale the supply side. But welcome to the show, Jamie. Thank you so much for having me, Matt. All right. Well, let's talk some AI. But first off, let's just set some context for the audience. Give us a quick overview of what is Mento. I kind of gave a little bit of a brief there, but what's Mento? What's the problem you're solving? How did Mento come to be?

1:12Yeah, absolutely. So Mento is a coaching and mentorship platform. The problem that we're solving is that we feel that in today's world, learning at work is broken. You will be shocked to know that companies spend$400 billion annually on learning and development at work. And yet almost every manager feels like they don't have enough support. All their ICs, every IC feels like they have probably virtually no support at work. And besides probably their manager and some peers. and that is not because companies don't mean well. They do. Every company knows that it's their responsibility to improve the performance of their employees and that is often through investing in them.

2:06But there just aren't that many good solutions out there. The really, the best solution out there is one-on-one coaching because it's personalized and in context. But even then, it's not as good as it needs to be because you might not actually be getting coached by someone that knows what you do or understands what success looks like in an organization or has led product somewhere. And so that is where we've started, which is all of our coaches at Mento are operators, meaning that they've had and have careers leading and working at some of the world's best companies. And we take a blended approach.

2:45So this is very unique to Mento is that we believe in the power of coaching and mentorship. Coaching helps you thin and mentorship helps you do. And I think that's one of the biggest challenges that we have is that people might be telling you, hey, you need to do this or hey, you need to do that. But how do you actually get there? And that's where support from people that have been there and done that or not be a sounding board and help you figure out is really where our magic is. yeah you're bringing back like some bad memories of having to take like the forced training back in my you know corporate days of being in large corporations and yeah you know the worst part about it is when you have this horrible training that the you know individual contributor the team members are not getting value from it's sucking their time and you're spending a bunch of money on it uh so yeah it's good training and you're like wow that was great i learned so much actually being able to go apply that because you're talking about well people are complex beings communication is challenging workplaces are challenging it's not as simple as let me apply this formula to a conversation with someone like it just that's proactive and not the thing are like learning is this thing that's way over here on your left and the actual practice and application is way over here on the right and they very rarely intersect and so that is the thing that we were constantly bringing together, both with our live coaching and now our AI-powered coaching product as well, which is bringing it to people in the flow of work based on their context.

4:19Quick break in the pod. If you're listening to this podcast, chances are you've been thinking about how to actually use AI inside your business. And that's exactly why we built the AI Opportunity Finder. It's a free tool that helps you uncover high impact, tailored AI use cases based on your business, your goals, your pain points, and your industry. No fluff, no generic use cases, just real ideas that fit your business and the ranked by ROI potential. It takes about three minutes to run, and it's like having your own personal AI strategist for free. If you want to try it for free, check out the link in the show notes or go to hatchworks.com backslash AI-opportunity-finder.

4:58Yeah, that makes a lot of sense. I think that, like I was saying in the intro, You know, these two-sided type of platforms and solutions are very difficult to get off the ground. Like I always remember the example with Uber when they were starting in a new city, they would just have drivers just drive around randomly. So there would be supply in case somebody needed a ride, even though the demand wasn't there yet. You had to kind of like force one side of it. But, you know, I love the model you're talking about with these coaches that are actual operators. But the constraint there obviously is there's just fewer of them relative to the people that could use their expertise.

5:36And I guess that's where this idea of, okay, well, how do you leverage AI in the solution to supplement these coaches becomes part of like how you're thinking about it. I'm curious, A, was this part of the original strategy? Has this manifested as of late as AI just kind of blew up and all these capabilities came to be? How has that evolved your strategy and kind of where do you see this going? Because right now you're kind of in this beta mode testing this out within the platform. Yeah, that's a great question. So when we started Mento, and this is actually one of your original questions, is why Mento and why did you start this business?

6:14When we started Mento about four years ago now, it was from our own experiences. My co-founder and I, he had amazing mentors and I was very developed for that. And I had actually had the privilege of getting access to a coach who was very much in the Mento model. She helped me think about my future and my career development while also helping me get better in the moment. And because she had actually done that before and I was like, this is so incredible. I'd recommended her to all my friends. and my friends were like, in what world do you think we can expense coaching? This was before the, you know, the better ups and all these coaching platforms really came to be.

6:49But even then, coaching platforms today, if it scales really well, which I don't think any of them do, because humans are expensive, no matter the quality at the end of the day, human time, people are expensive. They don't scale in the same way. And so it's not possible for every person in an organization to get the support they need to be successful. However, what we've been doing over the past few years is we, oh, and the reason why we started Mento is to democratize access to advice, guidance and support at work. Coaching and mentorship is the delivery or is sort of the mechanism of it, but we feel that it's in the best interest of people and in the best interest of companies if you give people the support that they need to be successful, right?

7:35And at the end of the day, most companies, there's not enough resources. Like companies just, there aren't great, there weren't great tools that aren't really expensive. So that's where we started with Mento. We knew that at the top of the market, starting at the top of the market, which was with one-on-one coaching, was going to be the place to start. If we can learn how we do this with people and we can encode that knowledge from a community of coaches who have the world's most valued expertise in their fields. And then we can encode that knowledge and build on top of our own proprietary mentorship and coaching frameworks.

8:12There would be a technology solution that allowed us to scale it. And so we started, you know, obviously AI. And one of the things that I learned from, I was at an Alphabet incubator called Jigsaw. And when I was there, we were working on an API that detected hate and harassment online. So we were already operating in the AI and ML space. It was so expensive. This was not technology. That would scale. You need humans to label every single piece of data. So when we started Mento, I knew that wasn't going to be the place that we could start. Like the technology just simply wasn't there unless you were like a huge, unless you had a huge knowledge faith already.

8:52You had a huge corpus of training data. You could train that data. So and then you would pay for it. And obviously, a lot of this technology, people are going to solve these problems. It was pretty obvious. Obviously, I did not predict the advent of OpenAI. However, it did feel that it wasn't going to cost millions of dollars for a company to be able to train a model like that just doesn't work. So we started with humans. We knew we would need the data. We knew that we needed to prove a different model. We did not believe in the traditional models that were out there in terms of pure play coaching or pure play mentorship.

9:25So we set out to invent a new one that actually drag performance at work. And now we're bringing that to people at scale through our AI product. Yeah. And for like listeners here, what Jamie was just talking about is a really interesting kind of strategic approach. Because a lot of folks right now, I think, are like, OK, AI is the thing. Let's just go straight into AI. Let's build the AI thing. Let's sell the AI thing. But I love how you've kind of taken the approach of, okay, let's figure this out with humans. And that gives us context for how to build a compelling AI solution. It's the same idea of if I'm like chatting with chat GPT, I got to give it context so I can get a good output.

10:06And I think a lot of folks are skipping that step right now, just in this AI hype. Yeah. So, you know, really interesting approach there. I think the democratizing effect too is you talked about that for coaching. I think AI today, that's kind of what's happened with these models being proliferated and open to all. They're just becoming cheaper and better. It feels like every other day something's going on there. And just to that point, I think the thing that we felt we solved with human coaches, we're not trying to solve with an AI-based coach, which is with humans. Humans that understand their context.

10:44As you well know, the context problem has not been solved with AI yet. Yes, you have to give open AI a ton of context, but being able to, and especially when you think about work, which needs to understand so many layers of context to be able to help you. Your organization, your manager, your web of relationships, your conversations, and those are changing all the time, day by day, minute by minute in many cases. and being able to have, being able to understand that context and then being able to actually then give that to everyone when they need it. Obviously with OpenAI and a lot of these other tools still, they're relatively primitive in being able to understand some of that information.

11:25And so that's really where we're focused in the same way that our coaches are. Because our coaches understand your context because they've been there before. They can help guide you. And that's the same principle that we're bringing to our AI product as well. Yeah, you hit on an interesting point. And this is another, I think, interesting thing to think of when you're building out a product is in this context, AI is complementary to the solution. It's not necessarily replacing the human coach because you would use them in different ways. And that's where I think it gets really interesting. If I'm, say I'm talking with some operator that works at Netflix or whatever through the platform, their time is very limited.

12:04I'm going to like talk about the most strategic, important things. but maybe I'm like you know I was doing this right before the session I'm facilitating a session around offer development like that'd be a great thing to ask a coach about but you may not use your time there so there you know becomes a valuable thing you could use the AI coach yeah use case right oh absolutely and I love that you brought up how it can expand where a coach can be because we you know even with Mento and most other coaching coaching experiences, it's every other week. That's twice a month. Like a lot going on in our lives and a lot going on in work, which is changing every single day.

12:46And another thing that's so interesting is that where AI and an AI coach can also play is that they can see your context. They can see what's going on for you. A human coach can't sit with you at work and see what's going on. So in some ways it can be there with you. You know, you're playing basketball and And it's like helping you call the shots, you know, while you're playing. That's very helpful. And then you can also watch game tape afterward. These things just haven't been possible before. And this is where AI is very exciting. Theoretically, yes, you could record something. But who's going to?

13:16I would love to think that we'd all record videos of ourselves all day and then just watch that and give ourselves feedback. That would not be as much fun. And that would certainly be very time consuming. So these are much more approachable ways for people to get the support, feedback, advice, guidance that they need. And I think this will completely change how we work. You won't need as much reactive tools in the workplace. Like performance management is completely reactive and like pretty much broken system from almost every company. it doesn't mean they don't play an important role but i think that the role that it plays is because that's the only way good way that we have right now is to think about people's performance is like after the fact and usually three to six months after which is pretty wild if you think about it yeah and you've forgotten everything by that point too and so i'm thinking through the solution i'd be curious how you're thinking about it and how you're thinking it's going to evolve because obviously there's like the idea of, okay, I am a human, I am a coach and I'm going to have this AI version of myself, kind of like a digital twin that, you know, thinks the way I do, has the history and experience that I do.

14:28And a mentee can talk to that specific coach. Is that the idea to where, and then like I'm getting into business model, I think this would be super interesting. Is it, how does it work then? Is it like a light, almost like you're licensing yourself out as this AI thing, or do you see a future where you're actually creating these AI coaches from scratch to where they necessarily, they aren't necessarily ever a human, but they are very tailored to specific use cases. Like, how are you thinking about it? Is it one or the other, both or something completely different? Yeah, I love that. So, and such a good question, because there's a lot of different ways where you could take this.

15:08So what we have now and what we've built is and building because it's ever evolving is kind of what we loosely think of as a mental expertise engine. This includes, we basically invented a methodology around coaching and mentorship in different coaching frameworks. These are the same ones that we use to teach our coaches. So we are teaching our AI coach things similar. We have collective expertise of all of our coaches. We have tons of different data that we've been able to collect at aggregate in terms of how we drive performance, biggest areas of feedback, different frameworks for supporting people.

15:44Then we have organization level data. So it's customized at an organization level. And then it's also customized at an individual level. So we think about this is sort of the mental expertise layer. And then on top of that, we have basically built like a context, a multilayer contents infrastructure to actually help pull that right piece of context. Make it useful and relevant to you at the right moment. With a relevant and specific UI so that you can actually deliver real time coaching and mentorship for you. So think about it more like that than, hey, I'm a person like we don't actually, at least today, we don't personally believe that AI is actually very good or LLM are actually very good at emulating people.

16:31They are really good at encoding expertise and helping you find that information. And what we're doing is building a multilayer system to actually enable to apply that to you at the right moment based on your specific context. And so pulling from all of that. So that's how we think about it. And at the top of that, from user experience, we've developed a framework for creating thousands of custom GPTs that are specific to you. So it's going to be a very interesting way. So you could have multiple agents, coaches, if you will, below the surface operating, helping you in a variety of different moments.

17:08A communication moment, a performance management moment, a coaching moment, a learning moment. Think of all these different moments that we have throughout the day and delivering a very specific one to you based on what we know about you, what's going on for you, what your growth areas are. Yeah. Oh, no, that's really cool. And it's you hit on something, too. Like, you no longer are kind of constricted to having one coach. You can have multiple coaches once you have AI. So back to the two sided marketplace, you kind of have, in essence, infinite supply, which is really cool. I think another thing I was just playing with the other day, I don't know if y 'all have tested it out yet.

17:47And a few of these models are using this, but like the ChatGPT, the app, it now has vision in essence to where you can do a couple of things. You can basically turn on the camera and it can see what you see, which is kind of an interesting thing if like you're in the flow of the event. Or you can share your screen as well. And it then has context for, you know, I'm looking at this email, I'm working on this workflow or anything. So I wonder, how do y 'all think of the multi-modality in the future? Because right now I'm assuming it's kind of just chat-based. It's kind of where a lot of things start.

18:25It's just chat-based, right? But you could go a lot deeper in a lot of different ways, right? Yeah, absolutely. So one of the things that were, so I think one of the big problems that we feel as well is that LLMs are still not as accessible as they could be. A lot of it does still depend on your prompting ability, the context that you did it, how you engage with it. And so building at the application layer, a few things that we think are really important in order to make this technology. And we think of LLMs as technology, not magical solutions that will solve all our problems. So maybe one day.

19:07And so it seems like. One day, yeah. Yeah. Like we can help do a lot of the prompting for you. We are doing that based on our interaction model. So you get context. We have a ton of context already. so we're actually able to do a lot of prompting behind the scenes for you so the ability to get great advice coaching mentorship support at work does not depend how good you are at prompting and then also another layer that is generative uis so can we pick the right framework for you the right is it chat is it voice is it then is it you know how are we interacting with you and i think also we're going to have a lot of other new modes of interaction that are invented or created out of this technology.

19:50Maybe it's a team-based interaction. There's so many different ways to think about modalities, which is where it also gets really exciting. Maybe it's an extension of yourself. Your coach can be, your knowledge will be encoded in that. Maybe you send that person, you send your coach over, your AI coach over, 20 team members to go kind of get some feedback from you, right? So there's so many different ways that we can interact collectively and individually with the technology that I think is going to be very exciting. And this is, I mean, one of the many reasons why we feel like this has the power to just be truly transformative for how we work because it's going to invent entirely new ways for us to work and collaborate together that I think will end up solving some of the hardest parts of work, which is communication, collaboration, managing yourself, working through difficult problems, loneliness, stress.

20:40I mean, all of the things that we don't have a lot of support around, but are the problems. Like I fully believe that most people like have great ways of upskilling themselves, learning about a topic, but very few ways to get sort of this workplace level support. So yeah, I love that you brought that up. I think it's a very exciting future that we will all live in. Yeah. You bring up an interesting point too. I hadn't really thought about it, but in essence, the coach, AI coach you're working with could technically be working with somebody else and have the context of what it's talked with you on and then which is super powerful because then it has context of the organization but i'm sure that also opens up interesting complexities too because you know they may have they may know some you know nuance kind of personal confidential thing yeah yeah which creates this whole new layer of and you know where do you draw the line and then yeah i hallucinate and all that kind of fun stuff yeah i know 100 and i think that's why you have to be careful.

21:40It's like we're not starting with that. The LLMs aren't yet intelligent. So it's pretty careful. Of course, we, number one for us is privacy and security. We definitely fall much more further along the cloud route because we are dealing in personal and potentially delicate human problem. Everything would be opt-in, right? And probably some level of customization, like, hey, I can train. I can tell them this is what I want my AI coach to know. but yeah you do have to be careful and I think actually it does remind me of sort of the human coaching wellness sometimes people ask us like you know I don't want to work with a coach that also like works with someone else on my team or a competitor or something yeah or a competitor something like that and then it's one of those things where you're like coaches can hold two things yeah at once and it actually not be a problem so we're kind of like okay if it ever becomes a problem, let us know.

22:37You can always switch coaches. And most people end up being like, okay, like it's fine, right? Because a coach can hold two things and has the sophistication to not bias them and, or not totally bias them, right? They can have a very private conversation. But that's a level that you would need to go to, I think, to be able to probably do it right. Or some variation of that where you could truly box off the information and the information that it has so that you don't accidentally, you know, share information. But I think all of that will figure in due time. Yeah, definitely. I'm curious, as you're looking at integrating AI into the solution, into the platform, like what's one of the biggest hurdles you face that, you know, listeners that may be looking at building their own thing or integrating it into their product, their business, what's like a hurdle that you all have faced or something that maybe was easier than you'd expect it?

23:28Oh, I like that. So I think there's two things. One, I think the initial, Well, this is specific to us, so you can tell me if this is not relevant. But when you are deeply integrating into a company like we are with our AI product, sitting in someone's Slack, for example, there we kind of thought, oh, hey, we're early. Like we have even partnerships in play with companies already through our live coaching experience. So we have our MSA terms and things like that. And so we were like, oh, you know, it's an early beta design partnership. like we'll be able to kind of bypass security concerns as long as we're not like training on data and things like that and like taking a very much privacy and security mindset.

24:10And that was not the case. Security teams were like, absolutely not. So we had to move more quickly on that. So I think if you're building and you do want to be more integrated into companies, definitely make sure that you're tackling that sooner rather than later. So that's one. And two, so that's one thing. And then honestly, though, So once you have, so for us, we entered in the SOC 2 type 2 observation window. But then once you have that, it's like pretty smooth sailing because you're not, we're not taking people's data and things like that. And I think also this idea that you need to collect a lot of data, especially at the outset of building, is not true.

24:48You know, just in terms of we're not going to be living in probably a world where you need like millions of different data sets in order to be able to provide value. some of that can come later it's going to provide your ability to build and test with people which is going to be your the most important thing at the beginning a lot of people will be like oh what's your unique training data this that and the other thing and or you know why aren't you collecting data when you're piloting with partners and you're just like that's just not really what's important right now so i think that was something that a lot of the engineers on our team felt really strongly about um and that's really proven to be true because it's enabled us to build along and wonderful partners.

25:22And at the end of the day, well, it's just a technology. And if it doesn't actually provide value for people or solve their problems, nothing really else matters. So figure that out first and then you can worry about some of the other stuff later. Yeah, the first pain point reminds me just talking to some of our own clients. Like you may have an existing MSA, but or whatever it may be, but their level of, it comes back to their risk tolerance and whatnot. But once you go into the AI route, which is something unknown, it does raise a level of questions and whatnot then you have the flip side where you know every company right now i feel like is being pressured to use ai in some compelling way whether it's the board or the executive team or whoever so it's okay how do i do it in a meaningful way as well and then the other point you mentioned just it's it's smart they're like the and i think this is something to like LLMs, they are extremely capable just out of the box.

26:19And I've gone back and forth on where you've heard of like the AI wrappers and you know, they're not differentiated. They don't have a moat and they'll just be killed by the next chat GPT feature that gets launched. But you know, if you can go back in time and there's so many SaaS products out there that are basically just a wrapper around Excel or some workflow or function. and making a lot of traction. So I think that's where I've come around in that belief where if you can target a very specific use case, industry, domain, workflow, and do it in a compelling user-centric way, there is value in it.

26:58And you don't have to go spend millions and billions to build a model, just use what's off the shelf and then you can supplement from there. Yeah, are you seeing attitudes for your clients about that? How do people feel about using off-the-shelf technology? Yeah. So I think in terms of the models themselves, you know, very open to using those directly off the shelf. So a couple of things. It's interesting question. There's obviously like the open AI and chat GPT's llama has gotten a lot of traction just because it's open source. The Metis created and then has been a really big thing for us for tree blogmented generation where you're kind of just giving AI context to your proprietary data.

27:39And the LLM is capable in and of itself. It's just missing the context. Once it adds the context of your business and your data, then it can just leverage that like it would leverage anything else. So that's been a very kind of powerful thing that we've dug into for different solutions. Yeah, that makes a ton of sense. And I think one of the philosophies that we have is that, hey, these foundational models are cutting edge. But it's also how you use them. and this is where really incredible so grateful to have work alongside such phenomenal AI engineers because they're really able it's not you know I'm just it's not just about having the technology it's always about how you apply it and how you make that accessible to people and there is so much room between an off-the-shelf you know just an open AI model and the layering that you can put on top of that to actually make that information useful to people there's so much to be done there yeah there's a ton of fine-tuning rag compounding systems that you can build and then eventually over time you know just because you start there doesn't mean you end there over time you know you transition and you fine-tune them and rag and train them on your own data and incorporate them as you learn and figure out what the value is but i think that is it's definitely it's one of those things that's like very busy and i feel like people feel like they have to ask that question in terms of is this just off the shelf?

29:02What is, you know, et cetera. I'm sure you all get that question all the time when you're in client meetings, but there's so much value to be derived at the end user point. And I know I kind of laugh sometimes because, you know, there's 50 great CRMs that are all huge businesses, for example. No one's, and yes, they're all a little bit different here and there and the other thing, and they all will differentiate. And, you know, they've all found their paths are different markets. And the same thing will happen with so many of these different applications that are being built and that's great and that's okay there's so much innovation to be had yeah that's a great stopping point that's a little compliment for the engineers there and i think too the other interesting thing i've kind of gotten into is like you in the past you had to have a path to scale some large thing because it was very expensive to build a product now it's not so you can build for a much smaller audience so much easier to get started, even if it's like a simple use case you're solving for yourself.