Scale Like a CEO

Why Psychological Safety Can't Be Rushed: AI, Cognitive Diversity & Team Collaboration with Nicole Radziwill

• Justin Reinert • Season 1 • Episode 49

Share episode

What if the real blocker to AI impact isn’t the model, but the space between your people? We sit down with Nicole Radziwill, CTO and Chief AI Officer of Team X AI, to unpack why organizations are sprinting toward a breaking point—and how a return to discipline can restore momentum without sacrificing speed. Nicole explains why technical debt is a strategic tool when paired with a plan, and how the obsession with shipping “now” quietly trades durability for theater. The result is familiar: stressed engineering teams, brittle systems, and roadmaps that mistake motion for progress.

Nicole takes us inside the origins of Team X—born from a decade of building cognitively diverse teams—and the creation of the Biodex, a practical way to reveal unspoken norms and hidden tensions. She shares patterns that stall collaboration, like the mismatch between how people prefer to send versus receive information, and misaligned response-time expectations that trigger needless conflict. From there, we walk through the Team X playbook built on the theory of constraints and three levers leaders can actually pull: calibrating friction, building just enough shared understanding, and tuning cognitive load to unlock flow.

We also dig into the AI shift. Prompting skills were a start, but many teams are stuck proving real outcomes. Nicole shows how Team X maps AI personas across a team to assign roles, close gaps, and define the exact needle to move—cycle time, defect rates, time-to-insight, or compliance risk—then certify return on AI with evidence leaders can trust. Along the way, we challenge Agile’s drift toward ticket velocity and explore how generative AI can disrupt low-value rituals while sharpening focus on customer value.

If you’re a founder, CTO, or team lead navigating fear, speed pressure, and AI expectations, this conversation offers a grounded blueprint: diagnose the space between people, prioritize the single constraint, and turn AI from experiments into measurable gains. Subscribe, share with a teammate who needs this, and leave a review to tell us the one constraint you’ll tackle first.

Nicole:

I think we're headed towards a breaking point. I think we're going to see that breaking point sometime, probably over the next 18 months. Debt, including technical debt, is a strategic thing that you should use wisely and you should use intentionally, right? There is absolutely nothing wrong with saying I'm going to incur some technical debt because I need to get to market. But then you have a plan for what's going to happen when we hit that point where we're going to have to resolve it to scale. Unfortunately, usually the strategy executive teams use is, well, the CTO is going to deal with that. We're just, we're going to do the business side. And when things need to change on the technical side, they'll deal with that. And unfortunately, that puts a lot of stress and pressure, not just on the CTO, but on the entire engineering staff. So when we do hit that breaking point, I think we're going to see a little bit of a return to the discipline.

Speaker:

In today's episode, Nicole Radziwill, CTO of Team X AI, explores how cognitive diversity can transform team collaboration with artificial intelligence. She addresses the critical tension between speed and quality in modern business, sharing insights on helping teams work more effectively together while navigating the challenges of AI integration in the workplace.

Justin:

Nicole, thank you so much for joining us on Scale Like a CEO. Just to get us started, if you wouldn't mind, give us a 90-second intro to you and your business.

Nicole:

Yeah, absolutely. So right now I'm the CTO and chief AI officer of Team X AI. And what that is is an AI-driven program. I'll explain a little bit more about that later, that helps groups of people work better together and with AI. And it helps them do that in the quickest amount of time because nobody's got the time for long training programs or, you know, sending people out to boot camps. Everyone needs to work at the speed of AI right now. And so what we've been working on for the past couple of years is to crack the code on how to get people to be able to do that more effectively. So the way that TMX works under the surface is by combining a technique from quality management, which is the professional discipline that I affiliate with most closely, called theory of constraints with some vintage AI. We call it vintage AI because it's, you know, very um traditional clustering algorithms, eigenvectors, you know, the things like the things that data scientists like myself get really excited about, but but you know, ordinary people don't. So what we wanted to do was make it possible for all, you know, there's a there's a uh a gold rush going on right now with AI, right? And AI, the the tools, the techniques are not quite commoditized yet. So what we wanted to do was make it possible to take all of the lessons that that we've gone through. I mean, you know, the first AI that I deployed for work was in 1998. We wanted to help people today take all of those lessons and really easily be able to apply them to getting people to work better together with each other and with AI. So a little bit about my background, you've probably gathered a little bit from the conversation already, but uh I am the technical co-founder. So my roots are in engineering. I started out as a meteorologist doing numerical weather prediction research. So we were doing big data and AI back in the 90s when nobody else was doing it.

Justin:

So, Nicole, that's great. You know, I'd love to hear a little bit more about what problem, you know, does your company solve and why does that matter right now?

Nicole:

So there's two really big issues that I see in business and enterprise right now. One of them is the addiction to speed, right? I mean, it's always kind of been there, right? Because companies have a uh an obligation to their shareholders to deliver immediate results and also to grow and scale. Um, but it's really gotten acute lately, meaning that, you know, I've worked with software engineering and IT teams for the past couple of decades. Uh, I was, you know, I was a software development manager for for quite a while. And there's much less patience for you know the discipline and the capability building because of the pressure to produce something instantaneously, especially in executive teams. And I've I've participated in executive teams that have this, that are dealing with this. There is an addiction to get things done absolutely right now, even if you cut corners, even if you stretch the truth to make it look like you achieved your goals. And to me, that really hurts because my my my professional discipline is quality and quality management, right? So, this addiction to speed, what it's done is it's it's systematically taken people away from that commitment to discipline, the willingness to go a little slower to get a solid, more scalable result. And so that kind of disturbs me. The other thing that I see going on right now is is fear. So here we are in 2025, end of 2025, and the past five years, past six years have been really uh fear-oriented. I mean, first we had the pandemic, and then people got back to work, and it we've had how many, how many rounds of layoffs have you heard about from companies that you follow? Right, it's just it's so many of them. And so there's this this underlying substrate of fear that that's really pervading everyone. I mean, it's part of it is is because of the many rounds of layoffs, part of it is because the economy, although it's doing well right now, it's it there's a hesitation that people have right there. There's a survival urge going on. You know, when you have a job, a lot of people just want to keep it and do whatever they can to keep it, even if it means being quiet about things that need to change. No problem. I'm keeping my job. Plus, now, since the arrival of democratized generative AI starting back in March of 2023, there's another fear, which is oh my God, is AI going to take my job? Should I even work on this? Because if I work on this, maybe my employer is gonna replace me because it's gonna be a lot cheaper. So, what I wanted to do was be a force in the world to kind of work on both of those issues at the same time. And that's one of the things that that are offering through TMX. One of the reasons why I'm really proud of it is because we help people maintain a commitment to those quality guardrails, like the things that you need to do with people and process and technology to maintain the changes. Plus, we also help everyone see that there's multiple roles for integrating AI into your business workflows. Like even the person who thinks AI is the worst possible thing ever and can only see the risks, that is a valuable perspective to add. So we wanted to help teams, work groups, be able to more effectively adopt those viewpoints into their day-to-day work.

Justin:

So interesting. There's there are so many things that you just kind of laid out that I want to unpack. One of them, I'm just curious, I want to dig in a little bit about you talked about this kind of discipline for quality. And it has me thinking about years ago when I used to teach a project leadership workshop, and we talked about speed to market and how, you know, it how getting your minimum viable product, even with errors, you still would get greater market share if you could be first to market and then fix it later. And I think we've seen so many versions of that in software and and whatnot. And I agree with you that we're pushing the boundaries of that right now. Like we just have to get it out there, get it out there. I'm I'd love to hear just a little bit more about that, given your proximity, like so close to this. Kind of how how often are we like pushing it too far in speed to market versus getting kind of the minimum viable product and the keyword there being viable?

Nicole:

Yeah, I think we're headed towards a breaking point. I think we're gonna see that breaking point sometime, probably over the next 18 months. Debt, including technical debt, is a strategic thing that you should use wisely and you should use intentionally, right? There is absolutely nothing wrong with saying I'm gonna incur some technical debt because I need to get to market. But then you have a plan for what's gonna happen when we we hit that point where we're gonna have to resolve it to scale. Unfortunately, usually the strategy executive teams use is well, the CTO is gonna deal with that. We're just we're gonna do the business side, and when things need to change on the technical side, they'll deal with that. And unfortunately, that puts a lot of stress and pressure not just on the CTO, but on the entire engineering staff. So when we do hit that breaking point, I think we're gonna see a little bit of a return to the discipline. Let me give you a specific example of that. The Agile movement, which started in the late 90s but really took off in 2008 with the launch of the Agile Alliance, it's been a long pattern in integrating agile development into IT organizations since then. But if you've been around since the beginning, a lot of us have noticed that it's kind of lost its soul, right? Instead of really being focused on what's the smallest amount of value that we can give to the customer, what's the least amount of work that we can do to get something meaningful to the customer? What's more, hey, I need to increase the velocity of putting out my Jira tickets. I need to write better user stories, right? We've we've drawn ourselves away from what that outcome is and the simplified understanding of that outcome. And our attention has been drawn to what the what are the tasks? Am I getting the tasks done in time or faster? And so I'm really excited about the role of generative AI in disrupting things, the Jira industrial complex, right? Because all of a sudden now it's gonna feel a little bit archaic to spend all that time writing tickets and to spend all that time going through that old archaic system when you could just have a better way to figure out what is the number one thing I can do right now? What is the highest value effort that I can do right now that I can can gather together my colleagues that we can collaborate on and get some meaningful value out. So that's why I'm excited about what I believe is going to be a pretty great disruption around that addiction to speed over the next 18 to maybe 24 months.

Justin:

Yeah. So tell me more about the work that you're doing to you know help people team more effectively and with AI. Tell me more about that.

Nicole:

Yeah, absolutely. So Team X emerged from another startup called Ultranots. Ultranots was launched in 2013 and it had a really unique mission. The idea was let's demonstrate that cognitive diversity can be a competitive advantage. And so when the company started as a professional services company in software quality and data quality, the idea was to build a workforce that was majority autistic, dyslexic, ADD, different cognitively, so that we could build a work environment that could be more ideal, right? We truly believe that you can build a workplace where everyone can thrive. And so along the course of the you know, 12-ish years since ultravauts was developed, you know, there's been a lot of positive things that have happened and a lot of negative things that happened. I mean, ultimately, the lesson there is when you have all of the best intentions as a leader and on a leadership team to really make a great workplace for everyone, why is it still not a great workplace for everyone? Why is it still so hard to lead people? And that's really what started us along the path to figure out, you know, is there a way to help a group of people work together more effectively in a fast amount of time? One of the things that you see from the academic literature is, well, just work on psychological safety. It's great. Psychological safety is great, right? You see teams that have it, the very rare teams that have it, you see them perform more effectively. But what we started to realize is that it takes an awful long time to build the fabric, to build the comfort and the trust between the team members so that psychological safety is a thing. Like it's not just a thing that you, as a leader or a founder, can give to your teams. You just can't say, we're gonna be a psychologically safe organization and have it happen, right? Doesn't happen. So what we discovered was that it takes an awful long time to get psychological safety. And then once you get it, it's really fragile. All it takes is one person coming in a team or one person leaving a team for the entire vibe to shift. So we started looking at that and saying, hmm, is there a way that we can use data, use what we know about people and their preferences and their work habits? Is there a way that we can figure out how to help them basically take the temperature of that space between the people so that they know what's the one thing we can work on together that'll help us work better together? And that's really the essence of the algorithms that are driving Team X under the surface. Did I answer your question?

Justin:

Yeah, I love that. And I am we're total, I'm just totally gonna go off script of the questions that we had planned because I just I want to dig in more because this is the work that I do, right? And I love this the idea that it takes a long time because it does. I had it makes me think of I was working with a group almost a year ago and they had severe trust issues and lack of psychological safety in the team. And so we talked about how to build that. And one of the ways we do that is with demonstrating empathy and things like that. And I had one person ask me, so how can I do this quickly? Like, how can I how can I how can I show my team that I'm have empathy with them really fast? I was like, Well, you can't like there you can't speed that up. We're talking about human biology and the way that we connect with others, and it takes repetitions, it is, you know, it is time after time, making sure that you're conscious and intentional about it. So what I'd love to hear are some examples of how what this looks like in practice. You know, so if I'm using Team X and you know, what's the if I'm a leader and I'm using Team X, what's the output or the input? And then how am I actioning that?

Nicole:

Yeah, absolutely. So the basis of Team X is that we recognized you're gonna be a lot more driven by your past experiences, and that's your past experiences in in life as well as in work. You're gonna be driven a lot more by that than by whatever work environment your employer is creating for you. Okay. So you number one thing is that your past shapes your present. We wanted to figure out a way to, okay, so you know, how can we diagnose? How can we figure out like what has this person's past been like and how has that influenced what makes them thrive at work today? We studied that for about 10 years, and we can't put this thing called the biodex. This is this is something that Ultra Nauts, the company, used for quite a while, and it became the basis of Team X. All the Biodex does is it asks you questions that are around your work styles, your preferences, that get to your beliefs, that get to the impact of your past experiences in work and in life on who you bring to your team right now. That's it. One of the reasons that we spent so much time coming up with the biodex questions is because, you know, working with a majority autistic workforce for so many years, what you recognize is that a lot of times people don't want to talk about their past experiences. Sometimes you can't even talk about your past experience. Maybe you're still figuring them out on your own. But those past experiences still impact how you show up for the people around you every day. So the point number one is we wanted to figure out a way to diagnose that. And that is the biodex. Then we thought, and this is actually the birth of Team X, right? That then we thought, okay, so now that we can kind of work with individual people to help them, to help them get a better handle on how they show up for people on their team. What happens when we look at multiple biodexes together? And that's where the really cool stuff started to happen because we realized it that's the magic is in the space between the people. It's that vibe that can shift when the one person comes in or the one person goes out. And so we started watching this. Um, and through we we probably did a year and a half, maybe two years worth of pilots with real enterprise clients. And so what we would do is we would go in, we would do the data analysis, apply the AI, and we would figure out ah, here's where we think the hot points are for this team. And then independently, we would have someone talk to the leaders and just get a sense for what the issues were in the team. And then we would compare them. And in 95, 96% of cases, we were spot on in what the hot point was for that team. And that's when we started paying attention, say, what's going on? Why is this working? To the AI question, because we were studying how to help cognitively diverse teams thrive in the workplace. It turns that what a majority autistic workplace needs to thrive are exactly the same things that we need to be able to collaborate with AI better. So that's where the crossover happened. That's where we realized, oh, because we can diagnose the space between people, we can also diagnose what those hooks are into how people are going to engage with AI, what they're gonna, what the their tendencies, the gaps, we could figure that out because of the work that we did in trying to make it more easily uh trying to trying to make more thrivable workplace that was cognitively diverse.

Justin:

So I love and I love the focus on cognitive diversity and on neurodivergent individuals and kind of helping them thrive in the work in the workforce. Uh I want to dig just another layer deeper. So again, let's say I'm a manager. What's the output that I'm getting that I'm gonna then shift the way that I'm interacting with my people?

Nicole:

Ah, okay. So let me step you through the way a T-MX engagement works. So step one is everyone on the team, I mean it when when we say the word team, it's just a group of people who have to get something done together, right? Some people don't really consider themselves a team, but they still have to get something done. They can still do teamx. So everyone in the team or in the work group fills out a biodex. You get a personal report for yourself that gives you a sense of of how our benchmarks and how our algorithms show that you're gonna show up pre-team. And then when we the next step we do is we analyze all of those biodexes together. We produce two things. One of them is called a team biodex, which is just a broadcast of the key work habits, preferences, the things that are gonna dictate unspoken norms or hidden tensions, the things people don't talk about because it's just so natural that they don't think they need to talk about. Here's an example of that. The number one thing that we notice when we look at the biotech data across teams is everyone has a preference for if I need to get a message out to you, I would much rather talk that message to you. I would I would speak that message to you. It's the easiest thing for me to do. But if I need to get information from you, overwhelmingly people prefer to get that information in writing. So all of a sudden, we have an imbalance, which is it's easier for me to communicate by talking. But oh, if I really want to understand you, it's easier for me to understand you if I get spit summaries or instructions or directions in writing, right? So the first step that we do, we do a one-hour guided discussion is just to make people aware of these things we don't talk about, just normal, ordinary things, like the way we handle our work day to day, can influence the tensions that arise between each other. Another great example is everyone has a set point for how quickly they respond to things like Slack messages or Teams messages, right? Some people are gonna respond within 15 minutes, they're on the spot. Some people do a lot more deep intensive work. And so they might take a day, maybe a little bit more than a day. So whenever you have those different tendencies in a team, you're gonna have tensions set up because the quick people are gonna be like, why is that person not getting back to me? Like, are they are they mad at me? Are they are they avoiding me? Are they even at work? Right. Whereas the person who is gonna have the long response time is gonna look at the person with the short response time as, man, they're really trying to just can't they just get any work done on their own? They're they're really trying to show us all up, right? And nobody talks about this stuff. So the first thing that we do in the guide discussion is just get people to to recognize that these unspoken norms and hidden tensions are at play. Then what we do is we have the second artifact that we provide is the Team X playbook. And all that is is that it guides the team in the direction of using theory of constraints. What is that number one area that they need to unblock to be able to collaborate more effectively together? So that's the human-to-human collaboration issue. And so that piece essentially focuses on three areas. So there's basically three different kinds of things you can do to unblock various collaboration. The first thing is change the amount of friction between the people and your team, right? Sometimes friction is good, right? Think about like regular regulatory compliance, right? Lots of rules, kind of stinks that you have to put all the effort in to please your regulatory agencies, but those rules are there for a purpose. They're there to add friction so that you don't end up doing something unsafe or putting a product out that's that's gonna hurt people or maybe be illegal. So friction can be a good thing, but too much friction and it's gonna cause conflict, it's gonna slow things down. So the first lever is friction. Second lever is shared understanding. And this is a big one with AI. The number one thing that's in the way of teams getting results is everybody having a common enough understanding of the problem to be able to work on that work process in a similar way. I've got a great story. This is this uh an engagement that I did probably 13, 14, 15, it was a long time ago. I did an engagement for social services. And the challenge that they were having was for placing kids in foster care. As soon as they got the word that they needed to place someone in foster care, it was taking them on average 157 days to do that placement. Obviously, a lot of dangerous things can happen in those 157 days. All we did was we took the five key people who were working on that process and we got them to diagram what is the process that you use to go from getting a new case to making a placement. Turns out that all these people who had been working together for five, six, ten years already, they were each doing their own thing with a view of the process that was slightly different from the way everyone else was looking at. All we had, I was really excited. I thought we were gonna do some fantastic math and simulation and operations research. Turns out we didn't have to do any of that. We just got them to literally draw one page, one process. They all hung it up next to the desks. And four months later, when we get back, when we went back to check and see if there was an improvement in the placement timing, it had gone from 157 days to 21, just by shared understanding. So lever number two, shared understanding. And then lever number three, which comes from the work with cognitively diverse teams, is cognitive load, right? It we all know that we get better work done when we're not overwhelmed with stuff, right? But at the same time, if there's not enough to get us going, it's kind of hard to get tasks done. So we need a little bit of cognitive pressure to make things interesting. Matter of fact, that's the root of a lot of the research by Tem Mahalley, who did the research on flow, but too much cognitive load and it's gonna, it's gonna inhibit our ability to get results. And so those are the three levers that we work with friction, which can be good or bad, shared understanding, which we need some, but not too much, because that's analysis paralysis. And then cognitive load, we need to reduce it, but not too much because we want to keep the work interesting.

Justin:

That's so great. I mean, anytime we can provide clarity for folks and teams around those things, I think that's where where teams start to thrive. So I love the work. I love the work that you're doing. And totally off script of our typical conversations, but I love kind of getting into it and learning a little bit more about it. Uh uh one kind of final question. What is what's the future look like for TMX? Where are you going? What are some new developments that you're working on?

Nicole:

So I I think what I'd like to focus on is the number one use that our customers have found for this, and that is the AI piece, right? So TMX started as a how do we get humans to work more effectively with each other super fast, right? Because nobody has the time to work on the psychological safety. When we realized that we were actually uncovering information that would help people engage with, particularly generative AI more effectively, we started to work with teams to build a language around what people's strengths were in engaging with the generative AI, what gaps existed, how you could use the distribution of those strengths and those gaps on a team to structure an initiative where everybody's being asked right now, hey, can you the CEOs are all saying, hey, we need you to show that you're integrating AI into your work processes. Show me the money, show me the benefit. You know, I see everybody else, whether it's true or false, everybody else is demonstrating the benefit. I want you to tell me how you're benefiting from generative AI. And, you know, unfortunately, a lot of teams are at a loss, right? So what's happened is they're learning how to do some better prompting. It's great. People are having fun, they're learning a lot, they're writing better, quicker emails. Reports are a little bit easier to do, status reports are a little bit easier to do. But uh most organizations are reporting that things are kind of things are kind of petering out, right? It's like and matter of fact, I just had somebody say this last night in a meeting with a client. He's like, we all know how to prompt, and that's great. But the question is like, you know, what what impact is it really making? We we can't we can't demonstrate the impact. So what we're doing by uh using the the AI personas that we diagnose in Team X as a jumping off point, we're helping teams figure out how do we organize ourselves? What what gaps do we have that maybe we need to get people from outside of our team, or maybe we need to just do a little more upfront planning so that we can do this. And then most importantly, this is the funnest part for me right now, helping teams certify their return on AI, right? And most of the time it's just clarifying what what needle exactly are you trying to move? Do we have the evidence? Can we justify and can we tell the story? And that's kind of fun. So we're using a tool that helps the that that helps to advance human AI collaboration. And we're using it to help people solve the very, very real problem of how do we demonstrate that we're actually moving important needles with what we've learned right now. So that's what we're gonna be doing in 2026. Yeah.

Justin:

I love that. I love that. Well, Nicole, thank you so much for the time today. I've loved the conversation. If folks want to get in touch with you, what's the best way to do so?

Nicole:

Best way to do is LinkedIn, always available there. We love talking about things that we've learned. Obviously, this has been a labor of love for a whole decade, and and now it's just coming to the point where we have a blueprint to offer to other people. Um, and you know, anything that you want to ask is great. Also, we we do offer referrals. So, you know, if if you're interested in in you know moving this into a company or you have a friend um who might want to put it into theirs, let us know because we do referrals.

Justin:

Great. Well, thank you so much, Nicole.

Nicole:

Thank you, Justin.