Michel Valstar - Founder, Co-CEO & Chief Science Officer - BlueSkeye AI

In this episode of the Full Stack Leader Podcast, host Ryan engages in a thought-provoking conversation with Michel Valstar, the founder and co-CEO of BlueSkeye AI. Michel discusses the innovative work BlueSkeye is doing in medically relevant face and voice analysis, particularly their clinical trial for detecting depression. He also shares insights into the ethical considerations and technological advancements that drive their work, emphasizing their commitment to privacy by running analyses on users' mobile phones. Michel reflects on his academic journey and how it has shaped his entrepreneurial endeavors, providing a unique perspective on the intersection of AI technology and mental health.

Transcript

Michel Valstar - Founder, Co-CEO & Chief Science Officer - BlueSkeye AI

Ryan: Hello everyone and welcome to this week's episode of the Full Stack Leader Podcast. This week I'm here with Michelle Val Star. He's the founder and chief scientific officer of Blue Sky ai.

We're excited to have you here, Michel.

Michelle: Thank you very much, Ryan. It's a pleasure to be here actually.

Ryan: So, Michelle, tell me a little bit about what Blue Sky does and how it actually works. 

Michelle: Sure. Blue Sky AI is a spin out from the University of Notumm. Started in 2019. And we do medically relevant face and voice analysis. And that means that we take a video and a voice recording of a person and we then recognize it the different facial muscle actions that you make, where you're looking, the tone of your voice, what you're saying, and all these behavior descriptors are then being used to recognize medically relevant behavior, including depression, anxiety, and fatigue.

We're currently running a clinical trial detecting depression in perinatal women. But in general, we have a technology that can recognize quite a number of different medically relevant behaviors. 

Ryan: And when people take those videos, are they happening on their phones or are they, do they go into an environment where there's specific cameras set up?

How does that work? 

Michelle: Yeah. There were two underlying principles for us. One is that we want, we are an ethical AI company. So we want everything to be as private as possible. And we want to make the biggest impact, positive impact to the world as possible. So we really targeting ubiquitously available hardware.

What's unique about the solution is that we run the face and voice analysis actually on edge on people's mobile phones and not on the cloud. And that means that it's private and readily available for as many people as possible. 

Ryan: So I know we've got a lot to talk about. With your current role in the company that you're overseeing. But I'd love to hear a little bit about where you came from and how you ended up in this position.

Michelle: Yeah, sure. I'll let you go actually way back. The funny thing is that my father was a psychiatrist who always said that he wanted to study electrical engineering and I ended up standing, studying electrical engineering in Delft and ended up building technology for mental health for the psychiatry business.

So we swapped roles over time. 

Ryan: That's great. That's ironic. 

Michelle: Yeah. So the way I basically did my, um, electrical engineering degree and my master's degree. And towards the end of that, I did a an elective module on artificial intelligence told by Hank, Professor Hank Coppola and 

 I really liked it. It was something that I'd never seen before anywhere. I think I really enjoyed the sort of the modeling of a problem, the modeling of the world, and then trying to optimize within that toy world. So I quite enjoyed the simplification, I suppose, of After that I was offered a PhD by one of copier's former PE students who now become a a very ambitious lecturer in delved my upon. And and I started working on face expression analysis. And that's never stopped after that. having done my first.

Did a research for my, actually towards the end of my master's degree, even before I started my PhD that already resulted in a conference paper. And it was my first conference paper and it was in Japan and I got to travel all the way to Japan and presented there that the IEEE conference of robots and human interaction, Roman.

Ryan: And what about what year was that? 

Michelle: This was 2004. The paper was so that I suppose that was also the time of the conference. And yeah, so that's 

Ryan: it's really early. 

Michelle: Yeah. 20 years ago at the time people were talking about, in the paragraphs about, you can use this.

potentially the future to do human machine interaction and to enable robots to actually see somebody's facial expressions and emotions. And you wrote that, that was the standard bit that you always blurb put at the beginning of a paper. But I didn't really think at the time about it becoming a reality and I never really considered that.

Ryan: And here we are. It is now a absolutely reality. And beyond that, we're actually using these facial expressions and voice analysis now to diagnose medical conditions from that. So we've gone there and beyond. Would you have guessed that it would have moved this fast within a 20 year period, or did you think that was a longer outlook?

Michelle: Not at all. Especially at the time, solutions weren't as data driven as they are now. So it seemed like every small improvement really required basically a completely different approach. methodology, completely different algorithm, a completely different approach to, to solve what was the next problem to do.

And, to, to the extent it's a bit sad now because what you need is the right team, the right data and the right hardware, and you can achieve. a lot. That, that's not to say that it's easy, but the formula for creating a successful technique is a lot more similar between, from one team to another.

Ryan: Yeah, that's amazing. So after you delivered that paper what ended up happening with your career? 

Michelle: I, yeah I came back and I was about to go on a year long trip around the world. But my, yeah, my supervisor convinced me instead to take on a PhD to cut back my month long sorry, my year long trip to a three months long trip.

And so I did that, started my PhD in Dallas and then then my supervisor called the offer to go to Imperial College in London and. That was great because in the UK it was only three years to do a PhD instead of four years. Imperial was very well it was a very prestigious institute.

And I'd seen Delft by then and I was really excited to see the big world also known as London. And yeah, so I took on the opportunity, had a really good time and, just really continued. enjoying this kind of research halfway through that really started to enjoy working with others on research as in doing collaborative research, sometimes me leading on a paper, but working with others, sometimes just contributing to others.

And I started to really learn the values of of, Networking and network effects, not just networking, so you get good introductions, et cetera, but also just the value of if I work with another teamand we have a paper, It's not just me who is going to root for that paper and recommend it and present it, but the other team is going to do the same thing.

So now I can double the attention to the paper. So things like that, plus, just a simple one. If you work multiple people, you just get. better outcomes. So I really started enjoying working with teams. 

Ryan: Did that kind of transition into you becoming an educator? 

Michelle: Sort of.

I mean, I suppose, along the same time I had to start doing teaching both individuals in their final year projects and some Of the classes. I started teaching some of the classes, the statistics class on machine learning, for example to students and I quite enjoyed that. But in a sense, that sort of teamwork didn't really come back until I started.

My own spin out company, but together with my two co founders, and that's when the, working together to solve a problem really started to build out again. 

Ryan: That's amazing. So really being able to apply all that knowledge at a later point and synthesize everything you had been doing over the years.

Michelle: I think it's remarkable how many of the traits of of an academic translate quite well into being an entrepreneur in the high tech sector, things like you have to be resilient and you have to really be able to deal with failure. That's something that is just second nature to academics because your papers get rejected all the time.

Your funding proposals gets rejected all the time and you just have to learn to live with it. You submit and forget basically that's a, that's the best way of dealing with it 

Ryan: It is really great. And you do a lot of presenting too, right? So you're constantly preparing for it.

Michelle: Being a successful academic is like being a one man business and you're constantly promoting yourself and you're constantly selling what your inventions were, what your research was, but also what your next proposal for funding, for example, is right. So you learn all those things.

There's other things, that are very different in academia from. The high tech startup business and academia, you don't move fast, right? And meetings in academia are really not good, not effective. So there's lots of things that I've learned since, but I think it's, it did set me up to become the kind of leader I am now.

Ryan: And so when you took that insight and that knowledge and began blue sky, and maybe you can tell us a little bit about blue sky, I you really have had to shift your mindset into kind of faster moving approaches, but still using some of the techniques. Correct? 

Michelle: Absolutely. Blue Sky AI was really created as a spin off from my lab.

It, we already had sold three licenses through the university to some of our technology, our face tracking and our facial muscle action detection technology. And I just really enjoyed helping these companies Fix some of their solutions and basically build better product. And that's not really pure research anymore.

That's just, incremental improvements, but I really enjoyed that. So when we created a startup from that, we basically took that knowledge, some of that technology, and we started looking for how we could start making the most impact with that. And that's when I really started getting to grips with things like.

Long term strategy visions a defensible, modes competitive advantage, as well as, how do you make money? Because all of a sudden we realized that it actually, it wasn't such a good idea or good thing that we could do it. Thousand different applications with our technology that always sounded like a good thing to me.

But then all of a sudden we were like we can only, build one or two products. So which one are we going to choose? That was a sort of a big shock to the system that actually all this opportunity was not a good thing. It was in a sense, great to have, but now we had to choose.

And how do you choose that? That was a hard thing. 

Ryan: That's it. It's almost like the experimentation moves from experimenting against AI itself and the growth of the actual science behind that into the science of customer application and like how it works. So you, the experimentation is a little bit different.

Michelle: To be honest, we did experiment with what markets worked and wouldn't work for us. So after COVID it's the first CES that was that was open again. we took six different markets that we want to explore. And we literally at a stand were.

Talking about these and just sounding out with potential customers and tech enthusiasts, but what they thought about that and we came home with some very clear learnings from that. And there was one area that I thought we would not we should not go in. And that was automotive.

And it turns out that on day one, the first hour of the, of that of that CES an automotive came to us and said you've got what we need and nobody else can do this. Please. Can you work with us? And within three months we were working with them as in, actually contracted to work for them.

Ryan: So why did you think you couldn't work with them in the first place? What was the concern? 

Michelle: My concern was that there were already other people working in that space that we. Didn't have a unique offering and that, automotive was too small of a business and they wouldn't be interested in what we thought was our, real unique selling point, which is medical grades, clinical grades.

recognition of medically relevant behavior. And in the sense we could have known that couldn't have known that, but it turns out that the, all the automotives around the time we were thinking about how to incorporate health and wellbeing into the cabin. And actually we were perfectly positioned to deliver that.

Ryan: So when that opportunity arises and you see a need to slightly pivot, doesn't sound like it was a huge product pivot, but a slight pivot, how do you realign the team or how did you rework the team to have everybody get behind that? 

Michelle: At first it was the making the decision that there were now two markets that we were going to to serve.

So that was the health market, which is basically customers in pharmaceutical, the pharmaceutical industry, and then the other one, the automotive. And that we, from now on, we're going to split all our activities in those two groups. So that, that was clear. We did health and automotive. So that, that was one.

And then it was yeah, question. We were raising funds at the time. So there was went to our CTO and we're like, okay we think we need these features. How many people do you need? And how long do you think it will take to get there? And then really rapidly building basically at the same time, Some table stakes features that we felt we must be in there as well as our unique selling points are unique features in there all at the same time, not trying to cannibalize our our health product roadmap.

Ryan: Had you signed any kind of an LOI or anything with the automotive company to, make you feel comfortable doing the R and D work, or did you guys Decide to jump in on SPAC and see what happened. 

Michelle: we had. It's a sort of half of both. We had this first company who were clearlygoing in the same direction we were, and they were happy to work with us on a three month time sort of contracts and we could see how that would lead into longer term, but there wasn't anything like a multi year contract.

Apple at the time. So partially it was a leap of faith. We of course were talking to other automotives as well. So we were hearing the same story and it became clear that there was a real thing to do here. But yeah we, it was also clear that, high volume revenues were still many years away.

Yeah. 

Ryan: Well, it brings up the question too, of when do you actually trigger more r and d on the product and how do you know which is the right thing to do r and d on? Yeah. Especially something that requires a lot of a lot of heavy lifting. So what's your perspective on that? Do you think the customer should lead that, or is that something No, your vision should lead.

 Yeah. 

Michelle: So I've got a very clear sort of view on that and that is that. I think if you're a true domain expert, I think you must trust your long term vision and you should use that to build competitive advantage, despite any criticism of the more short term thinkers around you, but then what you do is that you take that long term vision.

And that's what you're going to invest your R& D in. And you start saying, okay, that's where I want to be in five years time or in 10 years time. That's going to be a shock to the system, to anyone. What can I make as a minimal sort of bus stop that is still novel, it still requires research, but it's research that's on the way to that to that North star set of capabilities.

And then you start working on that. So if you have that, if you have that North star in terms of your research, right? So a lot of people talk about the North star in terms of the company vision and what you're trying to achieve in the market. But I think you can do the same thing for your search.

You can say what Technical capabilities do I want to have in 10 years time? And then you lay out that roadmap of how you get there and you allow the work that you do with customers and your immediate product needs to weave around that. That long term strategy, I think then you can get real synergies.

You, you either get situations where you meet a customer and they're asking for something and you're already halfway building it, or you get situations where. A customer asks for something and it's not yet on your roadmap now, and you can accelerate that work by bringing that forward, you're prepared, you have everything in place, you can put it in context, you get sort of network effects because a lot of our capabilities we build are.

Modular. So our gaze tracking, for example, is in the sense of discrete component. But if you now combine the gaze tracking with our emotion recognition, all of a sudden, we can do things like measuring your emotion. depending on what you look at, right? Whether you're looking at an object or a screen or a person around you.

So these two capabilities are now actually creating a network effect of creating more than one plus one outcomes. And, that does require you to have a long term vision and an idea of how you're going to solve some of these issues and being ready to, in the short term. make changes about your plans so that you can accelerate or make slightly adjacent features that will help ultimately with a long term roadmap.

Ryan: How have you chosen to split up work from ongoing like development and maintenance of the core product and then The R and D as a separate entity. Do you take the same team and work with it? Do you bring in another team? How, what's your approach to it? 

Michelle: Yeah. So this is something that took time for us to get to where we are now at the beginning, there wasn't so much distinction really.

And, we were a team of five in beginning and then about 10 but about. ago, we really started noticing that in a sense that didn't work anymore. And, And the real issue was specialization, I suppose, within the company, as well as being able to plan more and more deliverables in a.

way where we could be sure that we could deliver on time. So when we noticed that we started in particular separating products from engineering. Where product really sets the specifications, what is necessary? What is the, what are the priorities and engineering just delivers now within engineering, we do have the research, the R and D team.

So they have that sort of strange position where they both have to deliver to a product spec, but also have to keep up to date. The long term, it might. And the way we do that is that we basically we separate a portion of that time to to work on long term capabilities. Now, the other thing that we did is that we quite early on split off pure R& D from machine learning ops or advanced engineering, depending on how you want to call that.

But there's a different team that in a sense builds. The models the capabilities, and then there's a separate team that takes that and hardens those models, fine tunes it with more data and very crucially implemented so it can be deployed on edge because all of our all of our technologies is deployed actually on people's mobile phones or edge devices.

So there's a lot of engineering happens there that's goes beyond just being able to recognize facial expressions, but actually being able to deliver that then. On on complex hardware or demanding hardware. 

Ryan: Wow. Yeah, that's, that is a lot. And it sounds like there's a lot of coordination that has to happen between business and tech.

How much freedom do you give tech to come up with concepts and ideas that might turn into something, or is it something that is being driven from a business perspective and then having them experiment against? 

Michelle: I think we're setting goals in terms of accuracies that need to be achieved.

Matt or the camera positioning or the environments in which it must work. But then we do go to the engineering team and say here are your targets. Here are your constraints. Now go and solve this problem. And what we then do, we do ask them to come back with a plan in terms of resource requirements and timings in particular, and then do a bit of management with the exec team that do make edits that make sure we can deliver for our customers and hit any other strategic goals.

So it's a, it's almost like a an ABA process where we set specs, we set goals and constraints. We ask them to come up with a plan. And then we basically say, okay, that's great. But can we please, can we change the plan? A little bit here, a little bit there. What if you change this order and that order so we can deliver time.

And we do that on a on a quarterly half yearly and yearly basis, meaning we have three different sort of perspectives on ontime. And the level of goals are of course slightly different. 

Ryan: Yeah, that's really interesting. And what I was thinking about too, is that that coordination between the two is probably pretty important when you're dealing with areas that that your team is dealing with around facial recognition around a lot of kind of privacy oriented data.

How is it to manage? Groups in an area that receives a lot of publicity and maybe has a little bit of a sense of anxiety amongst the general public. How do you keep that in mind as you're working on these things? 

Michelle: firstly, we just talk about this with the team all the time and We brand ourselves and make it sure that everyone knows that we're an ethical AI company and we on purpose take the difficult, hard road to building this kind of technology in an ethical way.

And so when people join us, they, in a sense, know what they're signing up to. They know that we're hyper aware of the kinds of criticisms that are out there, but also that we have. That we believe that this technology ultimately is here to help people and that we understand the challenges, we understand the risks to privacy, the likes, and then we are going to work together to solve those problems the hard way.

And so when people join, they know that's what we're doing. And we talk about that. Repeatedly just today we were talking about, a new standard that we need to adhere to a new ISO standard. And everybody knows that adhering to these standards in a sense is painful because it takes a lot of effort and, can be quite boring to comply with all these things, but people know why we're doing it and that not only is it the right thing to do, we truly believe it is the right thing to build this technologies.

It also now puts us in a position where we are compliant and soon we'll be able to demonstrate that we comply with, for example, the new EU AI act and others won't because they haven't taken that route, and so it pays off. We talk about it regularly. We get people's feedback as well, right?

We sometimes get feedback. If people will say, Oh, why don't we just do it the easy way? And then we'll say if we were to do that we're basically breaching people's data privacy. So let's just build it the way we intended and run all the face and voice analysis on people's mobile phones and not upload the data.

Ryan: Yeah, that's great. So do you have a larger perspective organizationally around the ethics of this all meaning? Do you instill it in a way in your company's,core messaging both internally and externally, or is it something that you evolved over the course of time? 

Michelle: We have as one of our top missions to be the most trusted company to do face and voice analysis in the world Being an ethical AI company that really contributes to the lives and the health and the wellbeing of as many people as possible. That is our mission. That's our vision.

That's what we're going after. And in a sense all the rest is built around that. So yeah it's absolutely core to what we do.

what kind of advice would you give. A company coming into the AI space around planning their ethics approach and how they communicate it. 

 I suppose, you have to anticipate measure and mitigate. And that means that you need to start off with doing in a sense of risk analysis.

You you have to anticipate all the things that could go wrong. Before you even start full ideation of your solution. As you're, as you start thinking about your products, you have to think about what could go wrong and then you move on with building a set of measurements. So you can actually 

Make sure that you measure whether things are going the right way And then, of course, you mitigate. And when I say mitigate, I think especially when you start off, you have to be able to show how you anticipating problems actually results in architectural design choices and trade offs that you've made as a company.

For example, we build our technology as a two stage AI system where the first stage takes video and the way form in is very high dimensional and it outputs. On a frame by frame basis, what facial muscle actions you activated, what direction you were looking at, the tone of your voice, et cetera.

We call this behavior primitives, but that's the first stage. The second stage then separately does things like recognition of depression or the indication of depression or a mood assessment, et cetera. But by separating this architecturally in two pieces, We have now ended up with a situation where that second stage that does the pressure recognition, for example, doesn't need an image input anymore.

So we have now architecturally. enabled privacy and we can now deploy that first stage on people's mobile phones and we can deploy that second stage on the cloud, for example, and absolutely guarantee that this is private data and being able to actually go from, Hey, we're an ethical AI company to showing the general public and stakeholders in the space, how that translated into architectural designs is really powerful.

The other thing is trade offs. We don't work with the military. We don't work with law enforcement. And that is a trade off. We get lots of opportunities to work with these organizations. But if we were to do that, then we are really at risk. people being able to trust us, right? Because what if that is being used to reduce people's freedoms, for example, yeah, that, that is not something we want to be part of.

And even though it's in a sense, a lucrative business commercially it's a trade off That we have to make if we want to be consistent with our values and our ethical approach. And again, you, I think you have to be able to show that you make those trade offs. And not just, not just talk the talk.

You have to walk the walk and be able to show 

that. 

Ryan: Yeah, 

it makes sense. And I imagine figuring out how to communicate. The hard work you're doing in that area to customers is really important as well. 

Michelle: It 

is, and it we have now really translated this basically core, these core principles into benefits to our customers.

So it turns out that, of course lots of pharmaceutical companies are really interested in running studies to discover or validate. Novel biomarkers and they all, know that there's so much value in face and voice data, but they're very concerned now, of course, of doing that in a private and ethical way.

So voila there's blue sky all of a sudden. And we, we have turned our. Approach of doing things the hard way into an actual product where they can take some of our apps from our health platform, and they can distribute that to the participants and only the facial muscle action data, only that anonymized behavior data gets stored.

So all of a sudden, we have turned. what was a real obstacle for them to study face and voice biomarkers into a turnkey solution. And that took years for us to develop, but because we did this from the beginning in a principled manner, we basically, once we realized that there was a real need, we had the solution there.

And in a sense, all we needed to do is productize that.

Ryan: That's amazing. I'm really glad to hear that. This is the front of everything and that you're still able to keep growing in the midst of it. I think these are important topics for anyone moving into the space of AI, especially working with new and upcoming models. So great work. And thank you for sharing a lot of that.

We appreciate all of your insights. 

Michelle: No problem. 

Beak One 

Ryan: Michelle's mission has been clear to pioneer a new era of face and voice analysis while upholding the highest ethical standards. He sheds light on the importance of anticipating, measuring, and mitigating risks with a dedication to privacy and data protection. Thank you for your attention. These core values not only set the team at Blue Sky apart in their current market, but also paves the way for groundbreaking innovation in the intersection of AI and 

the human body. all right. Welcome back, everyone. Again, we're here with Michelle Valstar from blue sky AI. I've really enjoyed the conversation so far. And I'm wondering what your top five tips are going to be like. So maybe we can jump into those and hear what your thoughts are on leadership. How about we start with number 

Michelle: one?

Tip One

Michelle: My number one tip and I already said this, but. If you are a true domain expert, you have to trust your longterm vision and you have to use that to build a competitive advantage. Despite any criticism you get along the way from your short term thinkers. And I think that's really important.

Ryan: Yeah. That's pretty powerful in terms of like how you see the competitive advantage and then translate that to the team. Are there any specific examples or ideas around that? 

Michelle: Yeah, so I think it's a few things really that, that we see as the longterm vision. One is that people should be able to measure their own mental health themselves and be more empowered to do so. and secondly, that technology can do this more accurately than is currently possible with with self questionnaires and diagnostic tools that are currently available in mental health space. So that's what we believe. And the rest, in a sense, is a very long roadmap to get there.

The other thing, of course, that we believe is that this should be done, should and can be done in an ethical way, in a private way. And again, that is the hard way, but that's possible. And I think you should just work incredibly hard to make that possible. 

Tip Two

Ryan: All right. Amazing. How about tip number two?

Michelle: Yeah. So this is one that I've come to value recently a lot, but I think it's crucial for leaders throughout the company, not just the CEO or executives, but all leaders should really learn how to communicate. Clearly and concisely. I mean, it's one thing to create like a, a big competitive strategy, complete a tech roadmap, all your detail planning, etc.

You need to be able to be reductionist about that and have concise messaging that doesn't change with time. Planning changes, which will always occur, right? You will never deliver quite to plan. There will be changes when it comes to communicating to your team. You need to be able to be reductionist and simplify things and be able to communicate this long term vision.

Into, bite sized chunks that your audience, your team and also external audience can follow. And that's hard to do, especially when you are a true domain expert. It means that you need to sometimes slightly cringeworthily simplify things that you would normally go Oh, but actually it's not quite like that, but you have to simplify that.

And also it means that you have to learn when to shut up and not interject because and not to correct people, for example, because the details don't matter. It's the big picture that matters. 

Ryan: And of course have patience as people are trying to understand board. I'm sure that's a big one. 

Michelle: Yes, exactly.

I have been working in this area for 20 years. Obviously anyone who is new to this will, it will take time to, get that kind of level of understanding. And yes, you need to therefore be patient with people around you as they get familiar with the topic and your ways of thinking.

Tip Three

Ryan: All right, that's great. Let's jump into tip number three. 

Michelle: Yeah. One that I embraced from pretty early on, actually create a culture where failing is actually celebrated. And we do this every week on a Friday afternoon, we end the week with celebrations and we ask people to mention what went wrong.

Not only that, but they must say what went wrong and what they learned from that. And in the beginning we, everyone did that. But we're not too big. So we don't play a little game where you ultimately you thank someone and that's in the next person to go through their celebrations, which includes what they feel that, but because they don't know who's going to be asked next, everybody has to prepare in their heads what they feel that week and, failing.

Is super important. That's what you learn from. And if things are always easy you don't grow personally, but also the company will not grow. It's important that people don't hide that, that they talk about it, talk about what went wrong. And how they can avoid that in the future or what they learned from it.

Tip Four 

Ryan: That is a great tip. Let's go to number four. What do you have? 

Michelle: Yeah, so when it comes to being a supervisor to other people, a line manager, I think it's important that you're consistent with your one to ones and meaning that as you do them regularly use some structure in the way you ask questions and in particular I have started asking, and this was actually not my idea, it was the idea from our talent and culture team, but to ask about personal things, including workloads, relations, training and their own health.

But in particular, asking people about the relations within the company has really opened my eyes. Uh, there came questions or responses that said Oh I have difficulty working with this person because of that and that. And then you work that through and you go Oh, but that's because they don't, they're not invited to the same team meetings and they don't understand, they don't understand the context of where those questions come from.

Or they say, I've got a really good site relationship with a person X in, an adjacent. And that's really helping me understand why I'm building this this widget for this new product. So these are the kinds of things that are, people's relations within the company that as a manager, you quite often don't see, you're just too far removed from that, but they are, they give you real value and quite often early insights about things that go well and things that don't go well.

So I could really recommend people to ask about. What relations they have within the company. And what's changed. 

Ryan: Yeah. Team chemistry is incredibly important when trying to move towards pretty advanced goals. I'm sure in this particular case too, like having people collaborate in the smoothest possible way gets you to be able to develop this kind of complex software more effectively. 

Tip Five

Ryan: All right. Last one. Tip number five. What do you have? Yeah. Yeah, I think 

Michelle: meetings matter. but the quality of a meeting can be highly variable. And therefore, it's really worth putting time in it, reflecting on what makes meetings work for your team and what doesn't. And, honestly, I think a lot of this I've borrowed and stolen from Jeff Bezos and Amazon in particular, and some others.

I think it's incredibly important that you that you plan for meeting, that you set clear goals, what you're trying to achieve. I love the six page memo approach for when you're talking about complex matters, which in a deep tech company like ours, ours quite often happens. It allows people to really.

Make sure that they, that everyone is on the same page, literally, and then that they basically can rank their questions from important to non important. So that by the time that the hour long meeting or whatever, how much time you set yourself everyone has gone through the most important questions.

And if. And then in particular, it gives you also the opportunity because everyone has read the same thing. You can then let the most junior people speak first so that they don't feel influenced by the more senior people, which will always happen. So all of this is to say that meetings matter.

 make all your big decisions. in group meetings. So if you see the life of your company as a series of decisions, which are made in meetings, you better start putting some, paying some attention to that and making sure you get the most out of that. And don't just view it as a as a turn off live through it and and then forget about it forever.

Ryan: Great tip. And it definitely relates to number four as well. And making sure that people have a place to have a voice and you're also trying to get to the thing that's most important within it. So I really love your approach there. And I will also take that into consideration as well. It's great.

Michelle: It has some consequences because it basically means that you must make sure that you. Plan for yourself some time before and afterwards for, for preparing because you need to prepare for the meeting and then you need to take your actions afterwards. So yeah, I think there's a few, there's a number of people who said like a very busy diary it's not a recipe for success.

And I think part of that is is managing meetings properly. You need. You need to have time before and after to, to properly deal with that. 

Ryan: All right. That's great. Thank you for all of your amazing insights today. It's been really insightful and um, gained so many like great tips, but also a lot of perspective especially in the R and D world working in this burgeoning industry of AI.

We so appreciate everything you shared with us and thanks for being here today, Michelle. 

Michelle: Fantastic. Thank you very much for having me, Ryan. 

Segment Two 

Ryan: Michelle shared invaluable insights into effective leadership, emphasizing the importance of trusting your long term vision despite short term criticisms, particularly in leveraging technology for mental health solutions. This highlights the necessity for clear and concise communication, stresses the need to simplify complex ideas, and embraces failure and fosters culture where learning from mistakes, invaluable guidance for the success of any organization.