Hosts Eamon McErlean and Joe Devon interview Dylan Barrell, Chief Technology Officer of Deque Systems. The conversation explores his background, Deque's history and mission to automate accessibility, and the controversial keynote at Axe-con regarding the potential of AI to fully automate accessibility in the future. Barrell addresses concerns about job displacement within the accessibility field due to AI advancements, emphasizing the broader impact of AI across industries. He also shares insights on building a strong company culture and his experience writing a book on accessibility.
OUTLINE:
00:00 Opening Teaser
01:04 Introduction
04:46 Dylan's Journey into Accessibility
09:38 About Deque Systems / Automating Accessibility
12:04 Axe-con Keynote and Controversy Around Automation
24:22 Understanding the Criticism
25:48 How AI May Affect Developers & Other Positions
32:17 Maintaining A Culture As A Company Grows
41:36 Experience of Writing A Book
49:35 Future of Deque and Gen AI
55:17 Wrap Up
--
EPISODE LINKS:
Deque Systems
https://www.deque.com
Axe-con
https://www.deque.com/axe-con
Agile Accessibility Handbook: A Practical Guide to Accessible Software Development At Scale
https://amplifypublishinggroup.com/product/nonfiction/industries/technology/agile-accessibility-handbook-a-practical-guide-to-accessible-software-development-at-scale/
Extreme Programming Explained: Embrace Change
By Kent Beck, Cynthia Andres
The Ride of a Lifetime - By Robert Iger
https://www.penguinrandomhouse.com/books/556683/the-ride-of-a-lifetime-by-robert-iger/
Winners Dream: A Journey from Corner Store to Corner Office
By Bill McDermott with Joanne Gordon
https://www.simonandschuster.com/books/Winners-Dream/Bill-McDermott/9781476761084
Manus
https://manus.im
Dylan Barrell on LinkedIn
https://www.linkedin.com/in/dylanbarrell/
- There's an existential sort of angst that everybody who's a coder today has. That AI's gonna replace 'em and what is their job gonna be? So, accessibility is the least of it, to be honest, in terms of careers. The impact on all of us is gonna be tremendous. Of course, everybody's saying, "Well look at the horse and buggy and the cars, and there were way more, you know, every time something like this has happened, there've been way more jobs." I personally am a little bit more worried about this change just because of, it's the first time that we've ever had a change, which is gonna affect like 50, 80% of the jobs that are out there. The you know, the high end jobs. So, I think it's very potentially disruptive. And so, the fact that accessibility people are feeling this, it's like, join the club 'cause we've been doing, we've been feeling this for at least the last 18 months, right? It's not unique to accessibility at all.
- Welcome to Episode 8 of "Accessibility and Gen AI." A podcast that interviews the newsmakers and thought leaders in the world of accessibility and artificial intelligence. I'm Joe Devon, joined by my co-host Eamon McErlean. And today, we have the privilege of speaking with one of the most influential figures in digital accessibility, Dylan Barrell. He is the CTO of Deque, was ahead of the game in bringing AI to accessibility. He wrote a book about building an Agile Accessibility Program in the enterprise and even has a patent in the field of accessibility. Dylan, welcome to the pod.
- Thank you, thank you. Very good to be here.
- Glad we finally got to it. We've been talking about it for a minute.
- Yep.
- Why don't we just start with the background a little bit? Because you have, I think you were probably born in South Africa? You know, we've spoken quite a few times and then all of a sudden I finally understood your accent was South African, which I should have known. I do have an ex-wife from South Africa, so you know?
- Okay.
- Something I should have seen. But you've also lived in Switzerland, which I've spent a lot of time in. Puerto Rico, which I haven't. And I'm just wondering how did that impact your background? What have you learned from the different areas and what do you like about the different places you've lived?
- Yeah, that's a good question. So, I'd say the biggest thing I've learned about, you know, in the three or four moves I've made. Like, I used to live in Michigan as well, so I lived in the United States proper kind of, and then in Puerto Rico now, is that culture is one of the most, is one of the strongest things that drives people's behaviors and expectations. And culture's like water to a goldfish in that you don't realize that the culture is even there or what it is until you're out of it. And so, for example, you know, I went to live in Switzerland and there was just kinda small little things. And I was young, pretty young. I was 20 years old, so I was going out a lot. And you're meeting people and when you have a drink, you're supposed to cheers with everybody at the table. And you're supposed to take your glass, look the person in the eye, clink the glass together and say their name, right? And I'm terrible with names. So, this induced huge amounts of anxiety for me. And it's just a little example of if you don't do it, then you're perceived as being rude. And it just sort of made me hyper aware of culture and just how it's different from place to place and people don't even realize what the culture is. And they actually get, they get insulted if you do things that don't align with the culture without even realizing it. So, it's like the biggest takeaway.
- Yeah.
- For me. From all of those. Yeah.
- Yeah. I've learned that as well. Have you seen that, Eamon? In your background in terms of you have to look at the person, click, you know, clink everybody?
- Yeah, the cultural shift that I had was from Belfast, Northern Ireland in 1990 to Miami, Florida. So, to say it was slight cultural shock was putting up mildly. You know, blue sky, sunshine every day. white teeth, straight teeth they say is just, you know, it's-
- Yeah.
- It was a huge shift for us. It was. But to Dylan's point, there's these nuances. I think that's in every culture and it can take a while just to adjust and pick them up. Dylan, as Joe mentioned, listen, thank you so much for your time today. Appreciate how busy you are. Would you mind giving our listeners just an overview, an insight into how your journey began with accessibility and where that's taken you? Because you have a phenomenal background and a wealth of knowledge both from accessibility and business growth perspective. So, just any lessons learned along the way and what initiated that accessibility plan?
- Yeah, I'm a little bit unusual, I think. For a lot of people in accessibility is that I didn't really get into it because of sort of a personal connection. Although I do have a sister who has an intellectual disability. It's sort of, it's not something that entered my consciousness that I should personally get involved with that. So, I got involved because the chairman of the board at Deque sort of introduced me to Preety Kumar the CEO and thought that I was, had my own consulting company a little bit like Joe at times, right? And he wanted to introduce me to help Pree with sort of the product side of the business. And it seemed like an interesting technological sort of challenge. And so, that's how I got into it. I must say that it's taught me a lot because when I first got into accessibility, I used to feel so sorry for people with disabilities. I like, you know, I'd see like a blind person or a person in a wheelchair. And the disability became so conscious for me that it sort of overwhelmed every other aspect of my interaction with that person. I didn't even know what to start speaking to people with disabilities about. And so, what I've really received back from working in disability is to be able to see those people just for what they are, as just normal people trying to live their lives. And I've gained a lot of really good friendships and learned a lot from people like that that, that I would never have, I wouldn't have had any of those friendships. And so, I feel like accessibility has given me a lot back in return. I feel very grateful for that because it's allowed me to sort of just understand, What seems obvious is that people with disabilities just wanna participate like everybody else. They just wanna be kind of quote unquote "normal people" despite their disability.
- Yep. Completely agree. I don't think I've ever met an individual with a disability that wanted to be either have sympathy for or treated be-
- Yeah.
- Treated different.
- Yeah.
- It's not that way, it's just, it's not that way.
- It isn't, yeah.
- I have met a bunch of individuals with disabilities that I've learned so much from in so many, many different ways.
- [Dylan] Yeah.
- But yeah, they just want to have the same access, same feature set, and the same chances everybody else to do what they need to do, period.
- Yeah, I definitely had the same experience. I remember the first CSUN Assistive Technology Conference. There was a huge party and there were lots of people with disabilities. And you're right, I think when you first enter this space, you might see the disability rather than the person. And very quickly you get to meet the wide variety of individuals. It's such a cool industry that you're, you have just a complete mind shift.
- And I mean, for me that, you know, this actually gets back to, I think, one of the things I think is important about accessibility is, you know? Accessibility for me is successful when we eliminate it essentially as a discipline, right? And what I mean by that is, you know, it's great that a lot of people with disabilities are getting employment in accessibility domain. But what you really want is you want everything to be accessible from the minute they're born or from that minute a person gets a disability, to the point where they're not thinking about their disability, where they're thinking about the great things they wanna do, the things that inspire them. And so for me, you know? The goal of the accessibility, we call this digital equality, is really to get rid of our industry altogether that you know, I mean, it's never really gonna happen, but as much as we can make it happen to the point where it's sort of just part of the fabric and people with disabilities can dream of being anything, that's then we've achieved it.
- Completely agree. And that's what you said at the start there about stop thinking of it as an addition. That's one of the things that we're trying to do at ServiceNow. Yes, you can level up training and awareness and knowledge and skill sets and embed as much automation but the key is like, really embedding it into the requirements, features, phase, design, and stop thinking about it as an add-on. It's not an add-on.
- Yeah.
- It just should be part of the default definition of done. So, for those listeners that are not familiar with Deque in the background, would you mind just giving us an overview of what Deque does?
- Yeah, so Deque was founded in 1999 by Preety Kumar. And her vision from the start was really to quote unquote "automate accessibility." And I think we're gonna get into this topic a little bit later. But she always wanted to, she saw Deque as a software company first and foremost. And so, for a long time before I joined, that's what the Deque was trying to be, was a software company. She invented the category of automated accessibility analysis tool. She wrote the very first one and invented the category. And then what we discovered, and we might get into this a little bit more detail later, is we discovered that we had these tools to analyze applications and we were generating a lot of reports for organizations, analyzing 60,000 pages of the AT&T website in 2010, for example. And not a lot was actually happening. So, we started to look at why that was and there were two things. First of all, we needed to get to developers. So, we started developing developer tools. So, that was the first thing. So we, I think, invented, so you could say I invented the category of Developer Accessibility Tools. But then, we also realized that even that wasn't enough. We really needed to teach organizations how to embed it into their processes. So, we had to create a services organization. So, we have services now, but as a side effect of the software not really getting adopted and us realizing we had to provide services to help get the software adopted and get the change implemented within organizations. And then, we also realized that training was required as well. So, that's where our Deque University and our ILT in-person training arm came along. So, with the start as a software company, we've actually grown and the other, if you take everything together, the other two parts of the business that are sort of quote unquote "add-ons" are actually bigger than the software piece today. It's about 50/50 right now. So, that's sort of the history of the, you know, and our mission is digital equality. And so, we're trying to make it such that organizations produce digital experiences of every sort accessible from the get go.
- So, you alluded to something, let's call it the keynote. Can you share with us what is Axe-con? Talk a bit about the keynote. What the response is. And I'd love for you to steelman 'cause there was some criticism. I'd love to steelman the critics because I've tried to understand exactly what bothers people so much and your response to it.
- Sure. So, Axe-con started in 2020 as sort of, you know? CSUN was canceled, or it wasn't canceled but we didn't wanna go because as a company, we decided we weren't gonna go, which I think in retrospect was a good decision 'cause a lot of people got sick who went to CSUN that year. And we decided we were gonna create our own accessibility conference just online, right? So, that's how it started. I think it also started because we were a little bit not super happy with CSUN. It wasn't at that point. It's got a little better. But at that point it was not at all designed to facilitate the needs of people like us doing software and doing the services side of the business. So, we were sort of an add-on and it wasn't really well designed to help the people looking for us, sort of software and services and the people selling that. So, we also wanted to create a conference that was aimed at sustainable accessibility. So, when you look at it, it has three tracks. It has a designer track, a developer track, and then a wild card track, right? So, and that represents sort of our thinking about the importance of different aspects in terms of sustainable accessibility, right? You gotta start with design, you gotta make sure that it's embedded into development process, and then there's a lot of other stuff. So in the, there's, you know, a process change and that sort of stuff, which goes into the sustainable accessibility side of things as well. So that, it's our conference, we run it. And so, the keynote that we did this year, Preety felt very strongly. There's a couple of key trends going on in the industry right now. And one of them is EAA, and we think that represents a huge opportunity for the accessibility industry. It also represents a huge sort of stressor because it's gonna put a lot of pressure on the accessibility expertise in the world because there's gonna be a lot of demand, and there's not a lot of expertise out there. And so, that's the first thing. And the second thing is AI. And we really see AI, it's a wave. We can't influence it. It's coming. It's going to fundamentally change the way software is developed. It already is. And essentially what it's gonna do with the vibe coding and the implications of vibe coding. And for those people who don't know vibe coding, it's essentially where anyone can write applications even if they're not a developer. And the implications of what that means for organizations and how much more code and software there's going to be, we see anything from an accessibility point of view, which requires human manual involvement is going to hinder accessibility in quite a fundamental way that actually represents a threat to the opportunity of EAA. So, that's the sort of environment into which we're going. And what Preety was trying to say in the keynotes, in her very first slide, she put the slide up, it said, "At some point we're at 57% and we need to get to 100%," but she put 2025 above it. There was supposed to be a question mark there. There wasn't. And so, that was her very first opening slide. And it caused quite an uproar to the point where I think a lot of people didn't even listen to the rest of the presentation, especially not the part which, where I was speaking about. And so, there was a lot of activity in the chat with people getting incensed by this, right? So, that's sort of what happened. The steelman like, if you could say-
- By the way, before you go to the steelman-
- Yeah.
- Just wanna add a little bit of color. I wasn't watching it live. I had some other stuff going on. But I got texted right and left, like, "Are you watching this keynote?" So, there was definitely a lot of uproar in the industry. But I wanted to clarify something before you steelman. When you say 50, what was it 57%? 100%?
- Yep.
- Automated what? What exactly is going to 100%? Because I think that is super key that people are confused about.
- Yeah, so just put accessibility, right? And you know, that is our vision, right? Our vision at Deque is that we will eventually get to the point where close to 100% of accessibility, and I say close because, you know, getting-
- Accessibility testing or accessibility code? Like, that's what I'm trying-
- Accessibility development, coding, testing will be automated. That is our vision. I mean, if you take that threat of AI, right? Let's take that threat of AI. If the amount of code that each developer creates, if you do, you just put aside, set aside the concept of people who are not developers creating applications that are these complex applications. Put that aside for a minute, right? If you just take developers developing and the productivity gains that they get, if they're developing 10 or 100 times more code per developer, that means that the number of humans per line of code is going down, right? So now, if you need humans to test every line of code, but you don't have a lot of them, a couple of things happen. You start introducing even more friction into this. And then, so less of that accessibility gets done. So, the danger there is that if you don't bump up your level of accessibility to what I'd like to call the speed of AI, then we're putting the whole of accessibility at risk because companies are just not gonna slow down. There's too much competition risk from AI. There's too much opportunity risk from not doing things at the speed of AI that it threatens accessibility as a whole. So, my worst case scenario is the companies in the EU have this pressure of AI. They have the pressure of the EAA, and they fail miserably at doing things accessibly because of this, and they start putting pressure on their governments to delay the enforcement, to water down the enforcement. And eventually what happens is the EAA ends up getting watered down in the same way that the AOTA got watered down. I mean, AOTA came into effect in 2005. It was supposed to be final. All the deadlines were January 2025. And what effect does it has? Zero. And that's because there was no enforcement, it got watered down, it wasn't followed up on. And that threat exists for EAA in exactly the same way. It could end up being, to be going from this huge opportunity to have huge impact on accessibility, to being a nothing burger. And as an industry we will have lost a huge opportunity, right? So, that's the sort of environment that we live in. So, our strong belief is that the way we do this, I like to call this fighting fire with fire. If AI's gonna write all the code, then AI needs to help us test it for, and make sure that it's accessible. Now, I'll say make sure that it's accessible 'cause not all testing, right? Part of the solution is helping it to generate accessible code. I mean, there's more than one way to solve this particular problem, right? And there's also the possibility that AI may help with ATS and help ATS to be better even when the code or the application itself isn't accessible. So, there's that opportunity as well. There's multiple ways we can use AI to attack this problem. But if we don't use AI, if we see AI as a threat or if we see automation as a threat, then we risk actually hurting the people we're supposed to be serving, which are people with disabilities. Like, you know, we may at some point talk about overlays and that, like, I invented the word overlay because our product called Amaze was the first product to do overlay type stuff. Now, we did it in a different way. It wasn't automated. We didn't say this is automated. We don't generate them. We actually manually coded them, but they were overlays, right? And so, I've always, my attitude and Deque's attitude has always been if we can create an automatic solution to this problem, we should. Why? Because we're trying to put ourselves outta business. We're trying to serve the people with disabilities. And you know, it doesn't matter how we solve that problem, right? As long as we solve it properly. That's the difference between sort of the Deque approach and what some of the other people that have been doing overlays what their approach is. At Deque, we try to do it ethically in a lot of different ways. It has to be a proper solution. It cannot be sort of a ghettoization or a partial solution or an over-promise solution. But if we can achieve that, we absolutely need to try to achieve it because it's the best solution ultimately, right?
- And part of that proper solution, I'm assuming as you try to increase, we try to increase automation across the board, is ensuring that individuals with disabilities are consulted with and are part of that process the whole way through the product development lifecycle. 'Cause I do believe that is still a key piece independent of the increase in automation.
- Yes, absolutely. Absolutely. I mean, you know? And this is also another area because AI is creeping into this area as well, right? So, design is becoming more AI-assisted. So is user research and usability research is becoming more and more AI is becoming involved in that. So, we need to make sure that these tools themselves are accessible, right? So that people with disabilities can also be included in all of that research and that design thinking. So, right from the start all the way through the testing, people with disabilities need to be involved. And yeah, absolutely. That doesn't mean that the people with disabilities need to be doing all the work, but they need to be involved. In fact, we wanna make sure that people with disabilities are the users of some of these advanced tools, right? That's the way that we really achieve success is by making sure that as new jobs are created, new high paying jobs and skills for AI-based positions and things like that, that people with disabilities have the opportunity and are fully participating in that too. So, that's a big challenge for us, but that's really what we need to try to achieve.
- I'm gonna go back to the question of steelmanning your critics. And the reason that I'm doing that is when it comes to conflict and conflict resolution, very often the two sides are speaking about different things. And it's important when being criticized that you show that you understand what the criticism is. Because what I would love to see after this podcast, I'm sure a lot of people are gonna take a look, and I wouldn't like any misunderstanding to continue. If there's disagreement and people are upset, at least I want them to be upset about the right thing, if that makes sense.
- Yeah, sure.
- So from, you know what I mean? So, from your perspective, like, are they upset because of fear? Are they upset for other reasons? What do you think has caused that upset and respond to anything that you haven't already?
- Yeah, I would say part of the upset was misunderstanding, like bad presentation. Like, you know, if you attended our CSUN presentation, you've seen we did a completely different presentation addressing the same problem, right? So, part of it was we did a bad job with the presentation and getting our point across. So, I'll say that first and foremost. But let's just take the criticism just for the people that are even criticizing say, our goal, right? Of trying to go for 100% automation. There tended to be two types of criticisms. One was if you say that you can automate 100%, why should companies invest and why should people learn the skills, right? If you're gonna automate 100%. So, that was the one line of criticism. And the second line of criticism is, you know, I've invested a lot in this industry and you guys are putting me out of a job. And so, you know, that's a bad thing. How can you, you know, every time I use your product, it's helping to train your product. And so, my use of your product is putting us out of a job, right? So, those are the two sort of main lines of criticism. I don't know, Joe or Eamon if there was anything else that you heard that I haven't included in that.
- That's a really good question because I don't know, being so steeped in AI, it's, you know, every time that I'm looking at a new feature that comes out or at all these models, all I can think about is how would I automate every single aspect of what's going on in accessibility?
- Yeah.
- And I don't bother trying to build any of it for the simple reason that I know that you and a couple other companies have the resources to do it way better.
- Yep.
- And it's gonna come. But I just see that this is gonna happen whether we like it or not. And at the end of the day, and now I might get a bunch of criticism, but at the end of the day, I'm okay with, you know, if at the end of the day I don't have a job, but we've solved the problem for people with disabilities, then so be it. Personally, I think that the vibe coding, having done some of it, I just released a new little open source mini project. It's gonna introduce more bugs and I think there's actually gonna be in the short term, a lot more work for everybody to fix. Fix the vibe coding, do security analysis on it, do accessibility analysis for it. So, I think we're a ways off before it will solve it, but that's where it's gonna go eventually. So, it was hard for me to sort of understand some of the criticism. And I actually spoke to a lot of the people that were deep in AI. They all kinda saw your perspective. And the people that weren't deep in AI were the ones that were more critical.
- Yeah.
- So, I'm sure I'm gonna get a lot of crap to that, but.
- So, in terms of the criticisms, it's really those two things. So, let me address them. I'll address the second one first 'cause you kind of already addressed that. And so, you know, I think, for the last 18 months, I've personally have been looking at what AI is doing and how it's advancing. And sort of in the last, like about 12 months, 12, 9, not 12, sort of 6 to 9 months ago, it took a step change, right? And up until that I thought this was great, and this is really good. But about 9 months ago, I started to realize that this really could pretty soon replace all developers. I'm a developer, right? I'm a coder and a lot of coders work for me. And so, I think there's an existential sort of angst that everybody who's a coder today has. That AI's gonna replace them and what is their job gonna be? So, that's just reality. The other thing I will say is when I first started in my career, I remember hearing on the radio one day, some talking head talking about how in the future if people who've entered the workforce at that point in time, which is sort of late '80s, early '90s, is gonna have to relearn their career four times before they can retire, right? So, that's been a thing that sort of, you know, you've had to do even since the '80s or the '90s before computers were even sort of as prolific as they are today. So, I think what's happening with AI is it's accelerating that phenomenon, which is a sort of a phenomenon of the sort of economy that we live in today. I certainly, since I heard that that radio report back in the early '90s have at least learned five different careers since then, right? And so, it's definitely been true for me. And so, I think people that are saying, "Oh, I've spent a lot of time learning accessibility and this is gonna put me out of a job." I think that's an inevitability in most jobs in the economy that we live in. And so that's something we all have to live with. I have to live with it. They have to live with it. You know, it's just the reality. And so, the fact that AI is going to be, I think, way more disruptive, the fact that intelligence for the first time is gonna be everywhere, free, ubiquitous and intelligence that's sort of more intelligent than every human alive. That represents a fundamental change, a force for change that we, to be honest, I think if you think about it honestly, I don't think any of us can say we know what the effects are gonna be. They are potentially huge, disruptive, society-changing in potentially very destructive ways, forces, right? So, accessibility is the least of it, to be honest, in terms of careers. The impact on all of us is gonna be tremendous. Of course everybody's saying, "Well, look at the horse and buggy and the cars, and there were way more, you know, every time something like this has happened, there've been way more jobs." I personally am a little bit more worried about this change just because of it's the first time that we've ever had a change, which is gonna affect like 50, 80% of the jobs that are out there. The you know, the high end jobs. So, I think it's very potentially disruptive. And so, the fact that accessibility people are feeling this, it's like, join the club 'cause we've been doing, we've been feeling this for at least the last 18 months, right? It's not unique to accessibility at all.
- I'll just-
- It is not. It's no worry at all. Like, accessibility is one piece of it, but there's many, many jobs out there that have the same concerns. You think about content creators for creating new content. Like, you could name dozens of jobs that individuals are concerned. But it's that, I think it's that growth mentality. Like, you have to understand it is coming and make that decision, okay, do I learn more? Do I learn to try to broaden my skillsets? Do I have to reinvent myself? And I think some people will, and some people will be extremely concerned and hesitant to do so. Doubling back a little bit, because I did wanna ask more of a business-oriented question. You have phenomenal experience. You know, startups, bringing Deque. I think you've been in Deque about 15, 16 years.
- 15 years. Yeah.
- My question. Yeah, my question's more around seeing a business grow from that size and that volume from a culture perspective and mindset perspective. One of the toughest things I've ever seen is trying to keep that culture when you grow.
- Yeah.
- What's your advice there from a business and leadership perspective?
- Yeah, I mentioned early on that my big takeaway from moving to sort of four different countries or living in four different countries 'cause Puerto Rico, by the way, part of the United States, technically it's a completely different culture. It's like a different country. And in fact, sometimes each state is like a different country. I think you guys probably have experienced that too. Culture is everything, right? There's this book that says, "Culture eats strategy for business," right? And that is so true. So, I think at Deque we use a system called the Entrepreneurial Operating System. And it's not something we invented, it's something that we picked up from a company that does it. Gino Wickman is the inventor of it. And part of that entrepreneurial operating system is to be very thoughtful and systematic about your culture. So, they ask you to sit down and think very hard about what your values are. What are the values of the founders, and what do they want the values of the business to be? And then, codify those values, describe those values. And then, you hire people based on whether they have those values and you fire people based on whether they display those values every single day at work. And so at Deque we have our values. We're supposed to have five, we have six, which tells you a lot about Preety. Every time you ask her what are the top three things, three priorities, I shall give you five. It's my job as a CTO to figure out which of the three are actually our top three. But yeah, we've been pretty brutal. We've been pretty disciplined in making sure that people that don't fit the culture leave pretty quickly. We'll fire you as part of the EOS system, they say. You fire people much more quickly if they're not a culture match than if they are not in the right seat. So you say, "Right people, right seat." Right people means these people have the right culture. They're a cultural fit. Right seat means that they're in the correct job for them. They get the job, they want the job, and they have the capacity to do the job. And so, if somebody's a good values fit for Deque, but they're not in the right seat? We'll look for another seat for them. We'll look for another job for them where they are gonna be successful before we get rid of them, right? So, that's the best advice I can give to people is to do something like that. Be very disciplined and thoughtful and mindful about your culture. And it's not something you can just let grow. You have to cultivate it. You have to be very explicit about it.
- And do you believe that culture definitely comes from the top down? That leaders have to instill that culture themselves and live that the way that they're asking-
- Yeah.
- Their team members to live? Yeah.
- Oh yeah, if you don't live the, there's nothing more toxic for a company than saying these are our values and they're not living the values.
- I know, I know. Yeah. Yeah.
- Yeah, so-
- 'Cause I've seen that-
- It's important too.
- Oh yeah, absolutely. And it does. So, now you have to be as brutal upwards as you are downwards, right? It needs to apply to everybody, you know? And luckily at Deque, one of the values that Preety chose was open, direct, and respectful communication. So, what that means is open means, you know, we don't want people saying things behind each other's back. We want them saying them to each, to where that information needs to go. So, if you have a problem with me, Deque's values will say, you gotta come talk to me about that. If it's whether it's a decision I made, whether it's something I did, whatever it is, you gotta come talk to me. And the business benefit for that value is the information gets to the place where change can actually happen because of that information, right? So, if Preety's got a policy and people don't like the policy, if they go to Preety directly and they say, "I don't like this policy", she has the power to change it, right? And so it facilitates, first of all, it facilitates information getting to where it needs to get. And second of all, it eliminates politics because it's very explicitly a part of that value that if I complain to you about Joe, you are supposed to say to me, "Don't talk to me about it. Go talk to Joe about it," right? So, it stops office politics and that sort of stuff from coming up, which can be very toxic as well so. And I think that also, that value in and of itself is very, very important in helping that feedback get to people when they're not living the values, even the CEO. So, if I was gonna make any suggestion to any anybody out there. Pick that value, make that value part of your values, number one. And then, pick four more, whatever they are that are important to you. And because that value in and of itself is huge just in terms of getting just feedback about values first and foremost. Like, is everyone living up to the values? So yeah.
- I think there's a ton of wisdom in what you just said. I do believe that, speaking of controversy, some people are not gonna respond that well to parts of what you said. But I wanna bring that up because I used to struggle with having to fire people, dealing with some of the more difficult performers and realizing that they weren't that culture fit. And what really helped me deal with it was actually my brother told me a story that he had at his work where there was a leader who would not fire this person. Everybody knew this person was not a good fit and needed to go, but they just didn't wanna do it out of kindness until this leader's boss said, "Do you hate this person? Why are you not allowing them to move on to a career that they will do well at?"
- Exactly.
- And the question was, "Are you ever gonna promote them? Do they have any kind of future over here? Because if they don't and you're allowing them in this role, besides the fact that it's harming all these other people, they themselves are not gonna grow. Let them go to a company where they're actually gonna be a fit and they have a future ahead of them." And that's something that, you know, it's really about the frame in your mind of why you're doing what you're doing. And even the person that does get fired might come back. They might hate you, I've gone through this plenty, but they might come back in five years, which has not happened to me yet. And be like, "You know, it was the best thing that happened for me."
- Yeah, because it doesn't necessarily mean they're a bad person or they're bad at their job. It means that they're not the right culture. And I've seen it, you know, no names, but when I was at Apple, Apple brought in at one time a really, really senior leader in the retail side to replace a leader. And we realized inside two months, it was not a cultural fit. Period. It was definitely not a cultural fit. And inside four months, that person was let go. And it was a brave move because it was such a senior leadership. And I've seen it where that didn't happen. And it's amazing what can happen in the culture inside six, nine months.
- Yeah.
- When you have a leader, that's the culture that you believe in. It can completely change an entire company. It can. So, you have to pull that rip record if it's not the right fit, you know?
- Yeah, the most difficult people are the people that are actually otherwise very good at their job. So they're in the right seat, they're very good at their job, a lot of capacity, et cetera. Because those are the ones where the manager's like, "Wow, this guy's 10 times more productive than a lot of my other people." But they're a bad egg, right? So those, I found my experience has been those are the most difficult people to fire for values fit because they're so good at their job, but they're just creating such a toxic culture. And every time we've had somebody like that, and we've let them, ultimately let them go, as I always say, "Why did it take us so long to let that person go?" But those are definitely the most difficult ones.
- There are some where it's like, you really like the person or you actually think that they were great, but then the roles sort of had to shift as you scale to a certain point. Those are the really, like, heartbreaking because you have to be tough, you have to be firm. I don't mean tough, like mean. I just mean you have to be firm because you might have a certain relationship with them. They're not ready for what has to happen and those are the toughest.
- Yeah.
- But with that, I'm gonna change the topic a little bit and talk about books.
- Okay.
- So, I have written at least 20 books, but only page one.
- Oh really?
- Yeah. So, only page one. So, I have about 20 pages of a book written or 20 books written, but could never get that discipline to go all the way, which I hope that I'm gonna do this year. But you have written a book. Can you speak to the process of writing it and what the impact has been? And was it worth it? Because I had a friend who wrote a really great book and he took the only check for $150 that he got from that book and put it in a frame and put it on his wall. So, it isn't always a good business move, but I'm curious how it was for you.
- So, yeah, so Joe, your writing experience reminds me of smoking because, you know, I used to say, I think giving up smoking is the easiest thing I've ever done. I did it every day for five years. So, I understand from that perspective what you're talking about a little bit. So yeah, I mean, well, first of all, you know, the reason I wrote the book is because I saw the accessibility people who understood accessibility, but they didn't really understand development. And then, I saw the development people didn't understand accessibility, but they did understand development. And then, there were the managers in between who didn't really understand, certainly didn't understand accessibility, but what they didn't understand was what do they need to do to their organization facilitate to make accessibility sustainable, right? So, I wrote the book to really bridge the gap between those three people. I wanted developers to understand accessibility. I wanted accessibility to understand development. And in those two people, I wanted those two constituents to really understand how to apply their discipline of the other into their discipline, et cetera, right? And then I wanted managers to really understand the sort of resources they need to make available and what they need to pay attention to in terms of change, right? So, that's why I wrote the book. I didn't ever think I was gonna make any money from it. So, we give the book away for free. I have made a little bit of money. Like, I don't know, I haven't counted, but a couple of thousand dollars probably in total from royalties here and there. But really, I wrote it because I wanted to get that message out there, and I think there's been probably 10 times as many copies read that were given away than had been bought. So, I don't care about the money. Was it worthwhile? It was absolutely worthwhile because I think it's had a huge impact. I think I've heard people, they've come up to me and they've said, "You know, I read this book and I finally understood what I needed to do as an accessibility person to try to make accessibility sustainable." I mean, I've other people come up to me and said, "I read your book, I'm writing my own book, and I've taken parts of it. I hope you don't mind." So, you know, once you put your work out there, people learn from it and they hopefully make it better. And I think as an industry, I think it's had a little bit of an impact. Hopefully it's had a little bit of impact. And for me, that makes it worthwhile. How did I find the time to write it? The same way I gave up smoking. I didn't tell anybody. Rule number one, don't tell anybody when you're gonna do something big. For me anyway, this works for me. Don't tell 'em you're gonna do it. Just do it. And then, when it's done, you show them. So, I never made any promises to anybody. I just did it quietly, sort of little bit every day, a little bit every weekend. I didn't take time off of work. It was an hour here, an hour there. It shows a little bit in that the book is not very long. I didn't have a huge amount of time to do it. I wasn't paid in advance to do it. So, I kept it brief, a hundred pages, a hundred odd pages. And so, you can read it in like less than a day. But I think that's also sort the strength of the book is that it's short, it's not this huge tome. So, you can read it in just a day. And it was really, it was inspired by one of the most influential books I've ever read, which is a book called "Extreme Programming Explained" the first edition. And that book implements also sort of something that's become my philosophy. Like, one of the ideas that I find most interesting, that's not an accessibility related idea, but I apply it every day, is this concept of cargo cults where you have these tribes in the South Sea, you know, sorry, the Southern Pacific Islands where during the second World War there were these planes landing bringing cargo. And so, they developed this belief system that if they just put on things that looked like what we're all wearing here, you know, headphones and they performed certain rituals that these planes would come and drop boxes off full of cool stuff, right? So they thought, they didn't understand it, they just did these things and they expected an outcome, right? So, what the "Extreme Programming Explained" book did was it said here are the practices you can use to get extreme programming, which is what we now call Agile going. But more important than these practices are the principles. What are these practices supposed to apply? So, I try to apply the same sort of principle to my book. Here's what you're trying to achieve. Here's what we found works. But if this thing, if you try to apply this in your organization and this doesn't have this outcome? Well then, stop doing it or change it such that it does have the outcome. And if you can't achieve that, well then, don't carry on doing this thing if it's not having the impact. So, in other words, use your brain. And what I also wanted people to do is to sort of, you know, just think about, apply the Agile concept of retrospectives. What's working, what's not working, and come up with your own practices. So, I wanted to sort of take those principles and put them in a book. And so that's the approach that I took. So, hopefully it's not too bad an outcome.
- Sorry, I couldn't-
- I'm only a third of the way done. Sorry. Go ahead, Eamon. I did wanna know just who the authors were, but then go ahead, Eamon.
- Yeah, just, well, so we're talking about books and strategy and mindset and culture. I'd be remiss not to call out two books that I've read, and it's regards instilling a culture, a positive culture. Rob Iger, the ex-CEO of Disney wrote a book called "Ride of a Lifetime." Then Bill McDermott wrote a book called "Winners Dream: From Corner Store to Corner Office. And you talk about just a mindset of trying to embed a culture, like a proper culture. They were both, suggesting both. They're extremely, they're worth the time, they are. Sorry, Joe, go ahead.
- But yeah, I just wanted to know the authors of that Extreme Program. It was like Kent Beck or something?
- Yes. Kent Beck. And there was another author, but I can't off the top of my head. I'd have to go and look it up. But yeah, Kent Beck was the main author, yeah.
- Always love to credit whoever put in that effort, you know?
- Yeah, if Kent ever watches this podcast, which I hope you guys are-
- Hi, Kent.
- That famous. That he does. The question I have for him is why did he change that in the second edition? Because I thought the first, I think the first edition's better than the second edition, but, you know, that's just me. I'd love to hear his rationale.
- That's funny.
- Yeah.
- Wrapping up. Come back and wrapping up the Gen AI conversation. Is there anything either on Deque's roadmap or your personal roadmap from a Gen AI perspective that you're really looking forward to, or you're looking forward to it growing and seeing the benefit from it?
- Yeah, I think, I mean, you know, at Deque we started with Access with AI in 2018, 2019. And so, we started with convolutional neural networks for vision-based stuff. And we started with some OCR related stuff. And so, when we talk about AI at Deque, at least, for me, it's not just Gen AI. In fact, it started before Gen AI really became popular, right? So for me, I look, Gen AI just happens to be the, I wouldn't call it a fad, I definitely don't think it's a fad. It's just the most successful AI that we currently have. But it also has a lot of drawbacks and a lot of problems. And who knows? Those may get solved by tweaking Gen AI or they may get solved with completely different architectures or approaches. So, I'll just put that out there is that for me, it's not just Gen AI, it's all sorts of different AI. But I'm very excited about agents. I have been for some time. And I know that Joe is as well. And whether that's, the downsides of Gen AI is not predictable, you know, and it's not as precise, there's hallucinations. Whether it's hallucinations or sometimes the opposite of hallucinations. It's just stuff that gets omitted. So for example, when we were developing our LLM-based assistant product, sometimes you ask the AI, what's the difference between WCAG 2.1 and 2.2, and it'll just forget one success criteria, right? And so, that's sort of the opposite of hallucination, but it's a problem you still have to try to solve when you're dealing with Gen AI-based solutions. So, the reason I'm sort of excited about MCP and about standardized protocols for embedding sort of all sorts of tools into these agent infrastructures is that I think that that can help us solve a lot of these problems. And whether that turns out to be sort of a bandaid that eventually we throw away or whether that becomes something that becomes persistent and becomes sort of an ecosystem that we use to enhance these capabilities remains to be seen. I think it's likely to be the latter. I do think that MCP and that sort of standardized way of pulling in other tools is going to make these agent infrastructures more valuable than if they just based on the other, you know, the AI. Let's call it the LLMs today or the Gen AI today, right? So, I'm excited about that. And especially with respect to the opportunities that opens up for AI, for accessibility, right? Because it allows us to take this infrastructure that's being developed, that's gonna generate a lot of code and allows us to plug into it in creative ways to make it generate more accessible code or test the code that's generated. And so you know, I'm really, really hopeful and excited about that as being a way that as an industry we can leverage to help make better outcomes, at least on the generation and the testing side of things. Better outcomes for people with disabilities. So that's, I put that up there. Agent infrastructure with standardized protocols for calling tools as my number one. I'd like to hear what you guys have to say though.
- We can do 20 minutes on the topics that you've brought up here and we are running out of time. So, I'll just say quickly. Manus.im.
- Yeah,
- That's an unbelievable, like, there's just certain moments in AI especially-
- Yeah.
- The last six months, year, that you just feel like it's magic. And even though there's still some issues with scalability and a bit of polish on the Manus.im, that agent, it was like, it all for the first time really came together where you could just tell it what you want it to do and it'll go off and just take care of it. This is gonna be all over in the next probably few months, I would guess. We're gonna start to see more of these agents work. And then, MCP protocol. I've played with it a little bit. It was a little janky. So I said, all right, I'm gonna wait just a couple months.
- Yeah.
- Let's see if that gets better. So, I'm curious how you did there.
- Yeah, actually like you, you ended up using, I think you said to me that you used Cursor for that, right? Like, that's the worst.
- Yeah.
- It's like, you just happen to choose the worst infrastructure to test it in. So, go and test it with something like Goose or with something like Claude Code. Much better examples of how it can be implemented such that it's actually useful, yeah.
- Yeah, Claude is probably the best for it since they invented the protocol.
- Yeah. Yeah.
- You'd hope that they do it well. Gusto IO. Gusto IO is pretty good though, yeah.
- I'll check that one out 'cause I have not. Eamon, I don't know if there's anything else you wanted to add?
- No, I think we're good. Just honestly, Dylan, really appreciate your time, your insight-
- Yeah. Thank you.
- Your honesty, and direct feedback. Very, very much appreciate it. Thank you.
- Yeah, and I'll give you a chance to say the last word. But what I want to end with here is we could have done three hours easily 'cause I have a lot more questions. I'm sure Eamon does too, that we didn't cover. I mean, from Deque University.
- And Axe.
- Axecore.
- Yeah, exactly.
- It's at billion downloads. You know, the last I saw, which was 2023. God knows what you're at now. The impact that these things have had. So, we definitely should do this again sometime. And then finally, your son is a machine learning engineer at Deque, correct?
- Yeah. Yeah, that's right.
- So yeah. I'll let you take us out from there.
- Well, suffice it to say, I'm a very proud dad, right? No, he's actually been helping us with AI since he was an intern when he was at college. So, that's one of the reasons we've been it for so long is him because it was so difficult to find people who knew anything. So, thanks to him. He's taught me a lot. Him and some of the other people like Dylan Sheffer who've come in. And they have taught me a lot about this and inspired me and told me what to go look at so that I don't have to research everything myself. Yeah, well, thank you very much for the time. It's been a pleasure. Like, you guys should maybe do a Lex Fridman version where you talk for three hours of your podcast, right?
- I would love to. Believe it or not. I would love it.
- Yeah, no, it's been fun. Yeah, I'd like to talk. I'd actually like to hear what you guys think about some of the stuff as well more so. But yeah, it's been great. Thanks for inviting me. And if, you know, hopefully we continue the dialogue with our industry as a whole. And, you know, I'd love it if everybody in our industry had the attitude of let's fight this fire with fire, and let's figure out how we can use AI to push things forward. And because I think there's huge opportunities. Personally, like you said, personally for everyone who really grabs onto this with both hands and says, "Okay, let me figure this thing out." That personally, and then for digital equality. And for people with disabilities. So yeah, let's hope that everyone does that.