"Trust is the only thing that is really going to matter at a foundational level." —Brooke Hartley Moy, CEO of Infactory

Is your company rushing headlong into AI, only to find it's a 'square peg in a round hole'?

This week, Andrew tackles the critical issue of building trust in AI systems with Brooke Hartley Moy, CEO and co-founder of Infactory.

Brooke, with her experience at companies like Google and Samsung, cuts through the hype and reveals the biggest misconceptions businesses have about AI. We dive into the 'black box' problem, the importance of high-quality data, and why not all AI is created equal.

From seating Matthew McConaughey in the rain to high-stakes medical decisions, we explore the crucial role of domain expertise and the need to move beyond LLM-centric thinking. If you're an engineering leader grappling with AI implementation, this episode is your essential guide to building trustworthy and impactful AI systems.

Show Notes

Check out:

Follow the hosts:

Follow today's guest(s):

Referenced in today's show:

Transcript 

Andrew Zigler: 0:06

Welcome to Dev Interrupted. I'm your host, Andrew Zigler.

Ben Lloyd Pearson: 0:09

And I'm your host, Ben Lloyd Pearson. this week we're talking about the latest in a series of ThoughtWork memos on AI and the increasing divide between executive employees on their AI usage. But I think, Andrew, you had something you wanted to bring up first.

Andrew Zigler: 0:23

Oh yes. So a little birdie told me that you got a lot of attention on social media in the last week or two, and when you and I chatted about it, it seems like it happened when you didn't have any phone service.

Ben Lloyd Pearson: 0:34

Yeah, it was a, it was a little fun. Uh, so I posted this article about, Sam Altman saying that us copyright laws holding back ai. And, you know, I, I designed it to be a little inflammatory. I'm gonna admit that, but I still think I made a good point. it was funny 'cause I posted it right before I left on vacation and, checked my cell phone many hours later when it had hundreds of comments and of course had to turn off notifications but you know, my controversial opinions on. Copyright set aside. You know, I seem to have wade into like the front lines of this brewing conflict that I've seen, not just on my own page, but elsewhere on social media, and that is between creative professionals and software developers. So, you know, people that are in these creative lines of work, like designers, musicians, artists, they are really starting to witness a lot of the disruptive effects of ai. But unlike professions like software development, they don't really seem to be receiving a whole lot of the benefits that it brings along with it. So, you know, you think about software developers, like they really are experiencing a lot of like things that accelerate, that help improve quality, that help them learn new things. and the benefits have been way more substantial but you know, in, in the background we've also kind of covered in the past how like. Chinese companies are now entering this space and, you know, they have zero obligation to respect Western copyrights. So I kind of have every reason to believe that this disruption that creative professionals are feeling right now. Probably on only gonna continue to grow. Like even if Sam Altman and others can't train their models on copyrighted content. you know, I wore, I wore my fixed copyright shirt today.'cause this really matters. This is a matter I can talk way too long about and most people is, our eyes just glaze over. but I hope one thing we can't actually agree on is that today's copyright laws really do need to be rethought because I just simply don't think they're going to work in the AI era. Like the AI was never. Anticipated when copyright was created.

Andrew Zigler: 2:32

Isn't it funny how timing works? How you know you can post any other time while you're at work or dedicated to being online and you don't get much attention, but then you post something and yolo, throw your phone to the side, and you disconnect from the world and you come

Ben Lloyd Pearson: 2:45

Yeah.

Andrew Zigler: 2:46

everyone's pointing at you and you have all this unwanted

Ben Lloyd Pearson: 2:49

You know, and I thought I wasn't even, I thought I wasn't even gonna post this one and I just did it 'cause like why not? I was like, it's content that I wrote. Might as well post it.

Andrew Zigler: 2:58

Something you mentioned about, people in creative work, they're feeling the disruptive effects of ai, but they're not really feeling a lot of the benefits. I think this really underscores a lot of the disconnect right now that I see between techs and creatives.

Ben Lloyd Pearson: 3:11

Yeah.

Andrew Zigler: 3:11

someone in more of like a creative role who works. In tech, I feel like I'm often stuck between two opposing worldviews. A lot of the time when I talk about how I work or what I do, and I know many others, are in similar positions. And something I've noticed is that when we use AI for knowledge work in tech, you know, we say context is king, you know, that's what drives good copywriting, good code, good decision management

Ben Lloyd Pearson: 3:37

Mm-hmm.

Andrew Zigler: 3:38

iterations. But when you're using AI for creative work, cohesion. Is king. That means that the changes we want to apply with AI need to match the rest of the piece or what we had in mind for it. it's really hard to iterate and this is a completely different set of problems and solutions. You know, in painting is one of these ways that that technologists working with AI to creative work can change things, but there's still so much to tap into and explore. that cohesion ultimately is what relates that piece with everything else. you see a Van Gogh, you know it's Van Gogh because of the brush strokes, and you know, it's an impressionist painting because of how those brush strokes relate to each other. But since tech is really loud and proud about sharing their AI usage in the open, we're better at facilitating those strategies for context. But creatives, they're in a world where they're not talking much about that usage unless maybe they're dismissing it or disparaging it. There's a. Large polarizing view on it, and as a result, they don't share those strategies for cohesion. They aren't actually able to utilize AI as an artistic assistant in a parallel to how we can use it in tech. so that's just something that I've kind of been noodling on as I think about what is the root of these differences between how folks use these tools. Obviously there's other, concerns here involved in the conversations around copyright and the models that do diffusion and

Ben Lloyd Pearson: 4:59

Yeah, absolutely.

Andrew Zigler: 5:01

that are the root of a lot of issues with how artists view the usage of technology, and that goes back to your copyright. They're seeing a different world than we are.

Ben Lloyd Pearson: 5:09

and I have been wondering like, is this almost like a macrocosm of similar trends that we're seeing in software development, like some teams are really quick to embrace the benefits of ai. and I've, accelerated their work because of it while others, maybe that are doing things that are like tasks that are more like programming rather than software development. they're seeing more disruption and fewer of the benefits to the point where, you know, I don't think. AI is gonna take all software developers jobs, but there are many jobs within software development that it certainly will. And if your entire job is one of those things, then you may feel a type of disruption without the benefits of ai.

Andrew Zigler: 5:48

Very much, very

Ben Lloyd Pearson: 5:49

But speaking, speaking of AI disruption, let's talk about our next topic.'cause comes from a former guest on our show that I think has brilliant ideas and I always love reading her work. So what's this story about?

Andrew Zigler: 6:00

Oh yes. So this, this story is the latest memo actually in a series of memos from ThoughtWorks. this latest memo is from Birgitta Bockeler who, like you mentioned, was a guest on Dev Interrupted back in November. She talked about how we are making sense of a agentic ai. Birgitta is back with an article that explores the evolving role of developers as AI powered coding assistants become more capable and more prevalent. it points out how they can enhance productivity, but it requires constant oversight. And if you've been listening to Dev Interrupted, we've heard this again and again from leaders in the space that are using

Ben Lloyd Pearson: 6:35

Absolutely.

Andrew Zigler: 6:36

Creates so much opportunity, but it also creates so much disruption and it can really impact things like your time to commit your team flow and the maintainability of your code. You know, there are a lot of examples in here, in terms of actual, applied usage, that the author used as you went through different iterations of different tools and. What were the problems in her usual process that she encountered along the way? And it's really a hallmark of, of course, ThoughtWorks memos. I mean, it's very easy to understand, but it's also cut straight to the core of how an actual technologist would use these tools. Ben, do you wanna maybe talk about some of the things that stood out from the article to you?

Ben Lloyd Pearson: 7:14

Yeah, and, and I want to call out the, you know, our producer Adam mentioned, I, I believe this is the first guest that I ever interviewed on Dev Interrupted. So. topic definitely holds a place dear to my heart. But the thing that I really love about Brigitta is how she always focuses on very practical examples about the current state of ai. So this is like cutting edge, perspective on what AI is capable of. And why that's so important is because this stuff is shifting so rapidly. That I really think she is one of the best voices at keeping everyone in the know about the progress that AI development platforms are making. And there's a couple of points that she made that really stand out to me. And the first is by no stretch of personal imagination, will we have AI that writes 90% of our code autonomously in a year. So I love that really bold statement from her. She does also say that, you know, maybe it will assist in writing 90% of your code in some circumstances. she claims that about 80% of her development is AI assisted, which is like really awesome to hear. And I expect she's probably pretty far at, on the leading edge of, of that, just given the nature of her work. and one of the critical recommendations that I picked up from this that I. she recommends stepping away from GPT sessions when they start to feel out of control. Uh, and this sort of echoes my experience. Like most GPT sessions are ephemeral to me. I spin them up when I need them, and then the moment my problem is solved, I forget about them forever And so, you know, and I think really what we need to take away from this is that. engineering leaders, like you can't get complacent about AI assisted code. You need to maintain discipline around basic code quality standards. it's important that people are trained on those things and that they understand and agree with the practices that you're implementing.

Andrew Zigler: 9:00

Big plus one to the idea of throwing the chat away when it stops helping you or it stops

Ben Lloyd Pearson: 9:05

Yeah.

Andrew Zigler: 9:06

That's one of the biggest strategies and pieces of advice I've given to people always about using a tool like ai is that ultimately. If you're fighting it and it's not helping you step back, start over. And that's a really great observation here from her. it really kind of underscores how, you know, those LLM basics really carry you through this whole process. And everyone needs to be doing the upskilling work to really understand how this stuff works. so really great article. Really looking forward to the next memo on the topic.'cause I know it's an evolving subject.

Ben Lloyd Pearson: 9:34

Yeah, definitely. All, all the listeners need to head to the show notes, read this article. Lot of great advice. And I do also wanna mention that, we didn't get to cover it in this episode because it's a little bit bigger than what we were prepared to cover, but the, the ThoughtWorks technology radar, just came out to twice annual publication. always has lots of great insights about, the who's who of software development tools. so go check it out. Just came out at like, I think last week, So what's our, what's our last story for today, Andrew?

Andrew Zigler: 10:00

Our last story for today is looking deeper at the divide inside of companies, between executives and their own employees about how they perceive their AI usage and its impact on the work that they do. There were some really standout quotes and statistics from this article. there were a bunch of posts about it as well from a LinkedIn survey, that found that, you know, less than half of employees, 45% of them, think that their company's AI rollout in the last 12 months has been successful. And what makes

Ben Lloyd Pearson: 10:29

Wow.

Andrew Zigler: 10:29

powerful statistic is that that's. Compared to the 75% of C-Suite who thought that it went well. there's a big divide here between employees and how they're perceiving the usage of their AI tools within their company and the impact of it and their executives. only 57% of employees say that their company even has an AI strategy, but 88% of C-suite. Believe that their company do. So

Ben Lloyd Pearson: 10:56

Oh my gosh.

Andrew Zigler: 10:58

a misalignment between what's the reality within these companies.

Ben Lloyd Pearson: 11:02

so that's what about a 50% divide between employee perception and, and executives, like, that's huge.

Andrew Zigler: 11:08

Yeah, it's pretty stark, especially when you realize as part of this survey as well as it goes back to this 2024 LinkedIn report that said that 53% of employees were still hiding their AI usage from employers for the fear that would make them look replaceable. I. this really underscores how within large companies, small companies alike, there's still a really big disconnect between these two worlds. And there's still so much work to be done to bring them closer together with conversations about how this technology is actually impacting us. were there other things that stood out to you as well, Ben, in this article?

Ben Lloyd Pearson: 11:41

Yeah. Well, I love the hiding AI usage from your employer. You know, as somebody who formerly maintained a whole bunch of shadow IT infrastructure. You know, I can't officially condone that practice, but I do understand the motivations. But, you know, I really, it's, I like what I see is this like massive stark divide between employee and executive perception on ai. That frankly is shocking. In this article I read like there's pushback from employees, on AI specifically over a fear of being replaced. Like that's, that's, that's kind of gross. there's backlash against executives that claim that AI bots should be in the company org chart. Like, which, you know, let's just. Put this aside for a minute, like I think that's like a gross misunderstanding of how AI is actually going to work in the typical organization. Like it's not gonna be this all encompassing employee that does everything that a human can do.

Andrew Zigler: 12:33

Right.

Ben Lloyd Pearson: 12:33

I. But then there's also like employees sabotaging AI rollouts by just refusing to adopt it. My favorite quote from this was, it said, we're all sick of the effing chatbots. Don't ask me to use another chatbot. So,

Andrew Zigler: 12:47

that's pretty understandable. You

Ben Lloyd Pearson: 12:48

yeah.

Andrew Zigler: 12:49

getting shoved into a chat interface these days definitely adds a lot of fatigue.

Ben Lloyd Pearson: 12:53

yeah, I'll get to that in a minute. Chat is not a great in, it is an okay interface for ai. It's not the best, but in this survey actually, it applies to a pretty wide range of knowledge work, but I really think it is representative of something that we also encounter in the, engineering world, and that is unstructured AI adoption. So when you give your team AI tools without fully understanding their needs. Or providing training on how to use ai. You're setting yourself up for chaos, like what is described. In this article, like I mentioned, chat is not the best interface for ai. instead of chat bots, you should be thinking about how to deploy agents that are custom built for your workflows. So I call this like the tokenization of knowledge work. So focus on breaking work down into atomic units that are small enough for purpose-built AI agents to solve. With high precision and high accuracy, and of course you need guardrails on AI, too, there's a robust deterministic processes and tools that out there that come in handy. use CICD to enforce quality practices, implement static analysis, don't rely 100% on human judgment to maintain AI oversight.

Andrew Zigler: 14:03

Really well said. Especially about breaking work down into the small units. I know you've seen a lot of success with it. I have as well. That's one of my biggest piece of advice for folks right now is knowledge workers. Look at the things that you're doing, that you're repeating over and over again. Those

Ben Lloyd Pearson: 14:17

Yeah.

Andrew Zigler: 14:17

that you can solve with these ephemeral one-off types of solutions or chats. There's a lot of potential there for saving time.

Ben Lloyd Pearson: 14:24

Yep, exactly. So tell me about our guest today, Andrew.

Andrew Zigler: 14:27

Oh yes. So after the break, we're bringing Brooke Hartley Moy on the pod. She's the CEO and Co-founder of Infactory. And Brooke's work focuses on improving the trustworth worthiness of AI because after all, what good are tools if you can't predict how they'll work. So stick around for this fun chat. It's really gonna be a great one.

ben_2_04-04-2025_141354: 14:48

stop asking if AI can write code and start asking if it improves your workflow at LinearB. We've supercharged the developer experience with AI powered enhancements that cut through bottlenecks and save your team hours every week. LinearB now provides AI generated PR descriptions. To eliminate manual summaries and provide instant clarity, AI code reviews that catch issues early so your team can focus on what really matters And AI powered iteration summaries that turn retrospectives into automated data-driven actionable discussions. We're already using these features ourselves and the results are faster delivery, less cognitive load, and better development habits across the board. Wanna see it in action? Head to the show notes to schedule a demo, or join our upcoming live platform showcase. We hope to see you there.

Andrew Zigler: 15:35

today we're diving into one of the most critical challenges for engineering leaders, how to build accurate AI systems that actually work in the real world. Joining me is Brooke Hartley Moy, CEO and co founder of Infactory. Brooke has an impressive background spanning companies like Humane, Samsung, and Google. And now she's leading a company dedicated to solving one of the toughest problems in AI, trustworthiness. Brooke, welcome to the show.

Brooke Hartley Moy: 16:02

Hi, nice to be here.

Andrew Zigler: 16:04

It's so great to have you and we have lots to discuss. So let's go ahead and jump in. And the first thing I want to talk about is the speed of everything. AI is moving crazy fast and it feels like some businesses just aren't ready for it, Week after week, we're seeing new businesses and industries jump into LLMs, maybe without even having solid data practices or clear goals. Sometimes we don't even get a clear sense they understand what the LLM can or can't do. And I know you've seen this firsthand as well. What is the biggest misconception that you think companies have around AI right now?

Brooke Hartley Moy: 16:39

I mean, I think really been at the front lines of this. I certainly have in the roles I've played before Infactory and now in terms of Seeing companies trying to use LLMs to solve every single challenge that they might have as a business, and, you know, to use a trade expression, it's a little bit of a square peg into a round hole situation in some cases. You know, LLMs are this really magical technology. Like I'm fully AI optimist, excited about what LLMs can do, but they also aren't the end all be all solution or the silver bullet I think that people have come to treat them as, and not every single AI challenge, or even just business challenge, they get out of it. Not every business challenge should be solved directly with just your run of the mill LLM, or even a super powerful LLM.

Andrew Zigler: 17:30

And when you get down to it, brass tacks, what do you think an LLM is actually really good at for most businesses?

Brooke Hartley Moy: 17:36

It's in the name if you think about generative AI, which is really the kind of AI that we're talking about that LLMs power, you know, AI is sort of a catch all now, but we're really talking about one field within AI or within machine learning. Generative AI, LLMs are good at the generative side. They're very good at being creative. They're very good at content creation. We've seen that in, you know, models that produce images and videos and how much those have progressed in a short time. I'm sure many people have already experimented with things like email writing or blog posts, or,

Andrew Zigler: 18:08

Right. The

Brooke Hartley Moy: 18:08

New York Times just announced within the last 24 hours that they were going to use their own internal tool for helping with article creation, research, doing headlines, not going so far as to produce an entire article directly from AI, but to support that process. they're honestly a very great creative helper. if you want to produce something sort of net new and kind of go off into territory that feels unfamiliar and potentially coming straight from the artificial brain of an LLM, you're going to get pretty good results at this point from any LLM based tool.

Andrew Zigler: 18:42

I agree. It, it's really helpful for getting over the blank page and getting started with some kind of right, like some kind of critical mass to get you go. And sometimes you just need to see what you don't like to know what you do. And when you, you know, I think you and I can both agree that people playing around with AI and innovating with new things. There's no shortage of ideas or things that you could do to, change your workflows with the tool. But when you get the result from it, How can you really start to build trust in what you're getting out of your LLMs or your AIs? Because, you know, to us, it kind of just seems like a black box.

Brooke Hartley Moy: 19:16

Completely. And I mean, that's somewhat by design. And this kind of goes into the topic of, What are LLMs not good at? Which is they are inherently to the black box problem. They're not really that explainable. It's not very easy to understand where a decision came from. And in some ways you wouldn't necessarily want to. If you're thinking about the use cases that we just talked about, like, if you want the LM to give you that creativity, help you with the blank page in front of you, and to start to sort of spark ideation and all of that, you don't want to have it completely. Lockdown to only work with, you know, completely vetted, verified answers and obvious sort of delineation between, you know, source data and other things. when you're talking about other use cases that people are now increasingly trying to apply AI for, and particularly in the enterprise, and across much more critical use cases, that's where the black box problem Really becomes an issue. And that's where, at the end of the day, it's really about data. if you are going to trust the output of an LLM or the output of any AI tool, you want to know that it's being trained on the right data, that it is, if it's doing any kind of rag flow, any kind of additional grounding, that that data itself is trustworthy and verified. And you want to be able to have some exposure to like, what are those, decisions that it's being making behind the scenes in order to come to its answer. We're seeing this more and more, like, you know, Operator, the, you know, newest kind of forms of deep research that OpenAI has put out, and on the anthropic side, all of them are trying to expose more of at least the LLM's thinking. it still doesn't solve the problem completely of trust though, because. At the end of the day, the LM is still going to be inherently probabilistic in its response.

Andrew Zigler: 21:04

Right, exactly. So what I'm hearing is that the stakes and the quality of the data, if the stakes are high, then the quality of the data needs to be high. You know, I've heard this from other guests as well. And that seems to be a really big trend where when you're establishing trust, you know, if the stakes are high, the data quality, and the amount of information that's feeding into that decision making, it also needs to be high. So when you're talking about large enterprises, maybe ones that are typically really regulated, have a lot of restrictions. Maybe we don't typically view them as like a digital company. There's a very traditional enterprises that are sprawling. And you see them picking up these tools now every week. we talked about Goldman Sachs and how they adopted AI practices within their corporation. And so when you talk about the data quality, and being able to use the data. How can enterprises that don't typically have a strong data practice actually get good usage, out of LLMs with their data? Like, where do they start?

Brooke Hartley Moy: 22:09

Yes, it's a great question. And it's funny because it's a question that we've been asking for a really long time that certainly predates Anything related to LLMs or even, you know, predates people knowing what an LLM even was. Big data was sort of the, the moniker of its time. And I've been around the software industry long enough to have heard many conversations around how do we turn what is essentially this massive cost center of messy, siloed, not particularly well managed and clean data into something meaningful. And there's been various, Stages of technology that have come around to try to address that. I think the advantage we have now going into sort of the AI era to help these companies that you mentioned that are often more in the legacy side or have not even gone through the full evolution of, moving to the cloud or using things like data lake houses or the other solutions that have come to pass is. You don't necessarily need to, clean up every single piece of data in your current corpus. You can actually leverage AI, and this is something that we've been really focused on Infactory, to really understand, like, what data is worth investing in, like, what data Is going to answer the queries that are most meaningful to your business. Like where are the nuggets of gold hidden? And what is essentially like, you know, this desert of otherwise valuable, but maybe not so valuable information. And I think a big focus on that is then how do you set that up in such a way that the LLM, can understand that data and another application can understand that data, that you can use it, and get the information that you're looking for. That's a big focus on what we're doing Infactory, which is essentially on the data preparation side, making LLMs enterprise friendly by helping with this sort of data layer and addressing how do you make this data more useful in an AI context, recognizing that, like, there are huge advantages to that now. but you have to at least take that first step get them into that stage.

Andrew Zigler: 24:09

Right. And so not all data is made equal and some of it is going to be more important. I think that's a native instinct that most people have when they approach data, especially engineering data or things around Finances. Like, some things are just more important to look at and understand. And you really highlight, I think, for me, what I think is the anxiety when people think about the evolution of AI rapidly within large organizations, and, you know, you talk about like the big data, and being able to analyze what you have, and that was already. a problem that we had before AI got on the scene. You know, companies were already sitting on troves of data, more than they knew what to do with. And they were already going through processes of, you know, having the data warehouse, the data lake house, you know, go through all these different evolutions of how to access it. when you're talking about the nuggets of gold, and maybe you're somebody who's trying to find those nuggets of gold, obviously they tie back to your business, but how can you really start a data forward practice within your organization to make sure those nuggets of gold are always getting found?

Brooke Hartley Moy: 25:11

It doesn't require nearly the heavy lifting that I think people expect, I think part of it is recognizing that there are probably, a hundred different queries in a larger organization that covers 90 percent of what a business would actually want to know. And so when we talk about big data, I often kind of laugh a little bit about it because yes, there is a lot of data. corporations have more data than they know what to do with, and they're producing more, every second. And it's like The giant waste pile, like moving around the Pacific Ocean. And that's how I kind of pictured in my head of like all of this data that's just accumulating, accumulating,

Andrew Zigler: 25:49

Yes,

Brooke Hartley Moy: 25:50

mostly it's just, how do you get down to at a very granular, precise level? few things that really matter, and AI really does enable that in a way that we haven't been able to do before, you know, that's a lot of what we've been focused on over the last, few months is this idea that if you can understand just a hundred queries and get the right answers to those queries. You probably have made like a huge stride forward and being able to get the things that actually matter out of your data, rather than trying to sort of what has been historically the approach, because we haven't had this level of precision. We haven't had like scalpel level precision of trying to just turn everything into a data asset that it's not practical, nor is it really necessary. It's, it's a massive drain on time and cost and resources. And so. Businesses should be leveraging AI differently, and data engineers need to be thinking about this in terms of how do I actually focus on the things that matter? AI is really making that possible.

Andrew Zigler: 26:53

I think that's a big unlock because AI is really, at the end of the day, about focus. We talk about that a lot on this show, about what AI gives you is not necessarily the end result, But the ability to focus on the higher level things that the AI can't do. And the same thing has always been true about data. you draw a really good visual image, I think, of maybe having like a giant trash island of data. And you know, it's in the middle of the Atlantic ocean. It's the size of Texas.

Brooke Hartley Moy: 27:16

I realize it's not, it's, it's visceral, but maybe,

Andrew Zigler: 27:19

It's visceral, but so real. And in that environment, you know, that data, just like the trash that's kind of going in and out, it's cycling through. There's a lot of stuff coming in and out and you're building on top of it. And so really the key is if you're going to build on top of, what we think is like this, like layer in the ocean, you got to build fast. You got to be innovative. You got to look at what's there and be, inventive, but also decisive. if you're a team and you're trying to innovate. in that kind of environment, you have a lot of data, you have a lot of hypotheses, AI is giving you focus. How can you start iterating to get to the core of the problem we've talked about, which is trust?

Brooke Hartley Moy: 27:57

It's an interesting time for people in the development space. I always say to give another visual, it feels a little bit like building on quicksand because the pace is really fast, right? what was edge even just a few months ago already starts to feel obsolete pretty quickly. The ethics are constantly changing. we've got publishers suing some tools while other publishers are playing nicely and doing deals. We've got, DeepSeek. I think, you know, if you had told me that that was going to happen, it's like, that was definitely not in my 2025 bingo card. And so staying up to date with all of those things is really challenging. I think that the way that developers are going to succeed with the trust element is that first of all, Trust is the only thing that is really going to matter at a foundational level. In order for AI to have any meaningful impact on the enterprise and, you know, honestly, society at large, people have to believe that the outputs are, one, accurate and reliable, but two, that They're based on outcomes that ultimately have net benefit as opposed to, risks and dangers sort of embedded in the system. in my mind, like what I've seen succeed for people that are building in this space is just relentless focus on that idea. On that idea that you want to be working with high quality data, You want to be building with the mindset of how do I make this as reliable as possible for my end user? And I want to expose as much as possible for my end user. I want to make them sort of, pull back as much of the black box as possible. I feel like the winners in this space in a time of so much evolution and change are the ones that kind of are going to keep all of Those principles, top of mind in the trust arena.

Andrew Zigler: 29:49

I like the idea of building on quicksand, you know, it's going to be constantly shifting underneath you. It's a changing environment. And, and you've given some solid advice for how you can plan for things. You maybe can't even predict, you know, AI is moving really fast. We're trying to hit moving targets and build things that are not going to be obsolete in six months. And we don't want to make the wrong decisions that are going to cost our employers a lot of money or our, our team a lot of advantage. so there's a lot of, uh, you know, pressure, I think, and that pressure and trust, they work in a balance with each other because, you know, you want to trust as much as possible, but you also have to be able to keep up with. You know, competition. So it's a constant dance between them. And so when we get to this world where we're building with, LLMs and we're able to build trust from what we get out of them, I want to search here for, a really good insight for our audience that, you know, comes from like an engineering management Background. They're trying to implement their data practices, these data practices that you're talking about with AI workflows. A lot of teams right now have mandates from their organizations to create these workflows right now to experiment with AI. how can those engineering leaders partner with their non technical leaders and help them build that trust, but also help build impact?

Brooke Hartley Moy: 31:06

Yeah, I think this goes actually back to sort of the top of our conversation, which is where are LLMs effective and where they are not. And I think one area that has been. sort of deeply, underutilized as far as attention, uh, you know, has not gotten the primary focus is solutions and offerings that complement the LLM. And that I think business leaders and particularly non technical business leaders do see LLMs as sort of this panacea to anything that they want to throw at it. It's sort of on the engineering manager's to figure out how to compliment the LLM with other techniques and ML offerings that bring a level of determinism into the process. I think it's a, it's going to be a tall order to, to convince people that, you know, maybe we need to slow down the pace of, of LLM adoption. Uh, I think the genie is out of the bottle a little bit there. Now that people have some general understanding of what these technologies can do. But I think that there's a role for engineering managers to play in terms of pushing for features that support LLM usage in a safe and trustworthy way. And so that includes things like obviously evaluation. Um, I think evaluation is, is less and increasingly less important because it's, the models are becoming commoditized. How much one model is that different from another? But I think there are other features and efforts that engineering managers can take in the area of, you know, guardrails, obviously attribution and being able to bring in some sense of sort of data lineage and explainability. Anyone right now that's building in the explainability space, I think is on the right track and engineers looking at some of those tools and how they might leverage them. but I think it's, defining a full tech stack and architecture that isn't just LLM application. I think most engineering managers have certainly gone past that, that understanding at this point, but it's about, okay, what, what goes in between, RAG has been sort of the technique of choice, at least for the last year, now we're moving into much more agentic autonomous view of the world where it's not just about search and retrieval. It's about actual action. That's also going to depend on like the quality of the data that the agents retrieving, you can't trust an agent to go out and do something on your behalf. If you don't believe that it's accessing things that are ultimately correct and worth trusting. And so those are, those are the areas in which I feel like engineering managers have a role to play in terms of defining the safety and the trust that's going to be built into their AI systems going forward and to help bring business leaders along on that journey.

Andrew Zigler: 33:56

and so if you're in one of these engineering leaders and, if you're trying to start building these muscles, these are new muscles, you know, people haven't built in these ways before. you've given us some really good habits, I think, to start with, like around data practices. Are there any other habits that you'd recommend in this new AI native world?

Brooke Hartley Moy: 34:14

I think a few, one is, if you are building in the AI space, is, the builder, ultimately your responsibility to understand outcomes Your what, you know, basically everything comes down at this point to a query level. What outcomes are your queries driving? And so what I mean by that is that there's, we've sort of been talking about the, the first step in the process, right? Data quality, the foundation layer, what exactly are the services? And, you know, if it's a search service, if it's a bot, if it's an agent, if it's what, you know, application, whatever, what are the things that is supplying that foundation, both in the training of a model. The fine tuning, the grounding, all of that. now, particularly as we're moving, as I mentioned, into this much more autonomous agentic space where humans are unlikely to be always in the loop, at least, that's the dream, the vision we're all trying, we're all striving for at least, or at least some people are, some people I know are a little afraid of that. I think you also then need to understand. a very clear level, what are the steps that you've built up to that point? What actions are they going to induce? So I think something that's, you know, we've experienced this firsthand in some of my previous jobs is that we had a certain expectation that say, know, the, we had built, an AI agent to order food from DoorDash. And we had a certain expectation of how that would work. And, you know, you kind of started testing, you put into the world. a whole set of strange edge cases that you can't predict if you haven't done the work of sort of fully going end to end and understanding that it's like, oh, well, like what happens if, you know, the restaurant's closed? What happens if the driver, uh, doesn't show up the way that they're supposed to? What happens if this item is no longer on the menu? You know, all of those different scenarios. And I feel like something that like operating in this new world. Usually it's like you build an application, you take in data about its usage, you kind of understand each of those elements, you have some control over that, that like that whole playbook is completely out. And so something that I feel like people are sort of skipping right now that are in sort of the early stages is, Really taking the process end to end and like pausing to understand what are the outcomes that I'm actually driving towards and do I have then each of the steps in place to allow a successful outcome to happen.

Andrew Zigler: 36:47

That makes a lot of sense. Your example too makes me smile. The idea of having to innovate on all the edge cases when doing something like ordering a pizza. That's so true. But also,

Brooke Hartley Moy: 36:57

that I feel that acutely whenever I watch the ads for Salesforce's agent force where Matthew McConaughey is sitting in

Andrew Zigler: 37:05

yes,

Brooke Hartley Moy: 37:06

An agent didn't book his table. And I always think I'm like, yes, but obviously the hostess wouldn't have just sat him in the rain. This is actually an area where a human would do that more effectively than an AI, because a human would just see that it's raining. And

Andrew Zigler: 37:21

right.

Brooke Hartley Moy: 37:22

always, I'm like, that's a really great example in my mind of, of why agents are actually not quite ready for prime time and the work that you need to do to cover all of those scenarios, because if we were actually ready for the agent. To sit people at tables, we would have to plan for those things and they might not be very good at it relative to just hostess who can see that it's raining.

Andrew Zigler: 37:43

There's a lot of, there's a lot of hidden inputs that we take for granted as humans. Like we can glance outside. We can know it's raining. If you're like me, your elbow kind of creaks when it's raining. So like, you know when it's raining and so, but an LLM doesn't have a creaky elbow and it can't look outside. Um, it might be able to query the weather. Channel or otherwise get some weather data for you, but. Probably can't fully predict if it's raining. And, and also your example is great because it talks about stakes again. You know, we're talking about the stakes of what you're trying to do. That example, you know, brings a smile on my face about ordering a pizza and all the things that could go wrong. Well, what if you're trying to use an LLM during like, maybe it's used as part of like, Any kind of, like, medical booking process. Imagine it trying to make a decision or a flow otherwise around trying to get proper medical care. when you start implementing it into higher stakes workflows, you have to have done your homework and understand how things can go wrong. And I think that ultimately comes from domain expertise. Yeah,

Brooke Hartley Moy: 38:40

Completely. Well, this is an area too, where I, I often butt heads with others in the industry who take the viewpoint that, know, innovation for innovation's sake is always a good idea and that we should move as quickly as possible. And, you know, people who are not quite ready to embrace the AI revolution are to set in a certain worldview that it's about to be completely upended, all of these other things. And as I said, I am, I'm general, I am an AI optimist. I'm not afraid of robot overlords. I believe that this is ultimately going to be a net positive for humanity. But where I do buttheads with others in my industry and disagree is There are so many use cases right now where like, close enough is just not good enough. And you mentioned healthcare, you know, there, there are many others, finance, manufacturing, anything where the stakes are more than just, did your table get sat in the rain or did you order a pizza? And. I think a lot of folks have sort of looked at the LLMs as sort of this magical tool and have had positive experience with LLMs and see that the models are improving and see that they're getting closer and closer to perfect accuracy, but they're not at perfect accuracy. And, you know, within that, Whatever it is, you know, even if it's 0. 1%, it's certainly not a 0. 1%. I mean, I think for many things there, it's closer to more like, you know, 20 percent of, of, of the time you're not quite getting a high quality answer, even if it's not a full hallucination, that's problematic. The space for edge cases is then just significant and gets exponential. And the stakes are just too high. And, we talked to a lot of enterprises who. Would benefit from the use of AI, but can't take the full risk of it's completely autonomous and all, all decisions are being driven out of an AI system. There, there has to be a range. And, you know, again, going back to that New York Times example, I mean, they, they clearly have decided that it's worth augmenting the performance of their, you know, journalists, but it is not worth creating all their articles with AI because there's still a degree in which human expertise is going to matter.

Andrew Zigler: 40:49

I completely agree. It's about understanding the levels and also the nuances within your business and how to best leverage it, but also how to get the most reliable results. And that's why I'm so grateful for this conversation because you've given us such a really like Thorough Playbook. An understanding at the beginning how data impacts trust and how you can build impact through your systems by understanding what goes into them. You know, we talked a bit about building and uncertainty, being able to innovate. And ultimately, you know, we keep circling this point of talking about the stakes, uh, which is really, really great. It's really powerful takeaway that, you know, in our audience, I know a lot of our listeners, they really resonate with this. We're talking about folks who work in engineering industries that are in scenarios where, you know, everything needs to be taken into account. And if something goes wrong, those stakes are really big. So when you're weighing, you know, all of the pressure of implementing AI in this scenario, it's really great to get, you know, perspectives like yours and kind of understand how other teams are doing it. before we wrap up, I know you were talking about some of your opinions on AI and where we're going as a world, and where can our audience, where can they go to learn more about you and InFactory?

Brooke Hartley Moy: 42:02

Yeah, so infactory.ai would be the first place to start, if you message us, uh, I, I see everything that comes through. So even though, um, you know, we're, we're working with large enterprises, I, I take a direct interest in particular with what the developer community is doing and what engineers are working on. And, you know, we, we've really built. entire product with developers top of mind, because we've lived that, right? Like we've, we've seen the challenges of what it takes to build high stakes AI. The challenges that come from business leaders who think that, uh, LMs can solve all their problems. And we really want to help, folks that are building in this space who, who believe that there's a way to do this. That's. Trustworthy and effective. So, I love engaging with, with people that are building cool and interesting things who want to build it in a way that feels inherently safe and important.

Andrew Zigler: 42:54

Awesome. We'll make sure we get that link in the show notes for our listeners. I also recommend to our listeners check out Brooke on LinkedIn. I've been following Brooke for a little while. I love her posts. They're really, really fun and insightful. And just like this conversation, you know, Brooke always talks like a human in an AI world. So it's great to get a perspective, that's like deeply resonant with what's going on. That's it for today's episode. And to you, our loyal listener, thanks for making it this far. If you enjoyed this episode, please subscribe, share, you know, Send this to anyone you may know who could get something out of today's episode. We'd also love to hear from you on socials. So come bug us, come poke us, come leave us a nice comment. I'd love to hear what y'all are thinking. And see you next time.