From Front-End Developer to Accessibility Champion

Crystal Preston-Watson takes us on her fascinating journey from self-taught coder and journalism dropout to becoming a Senior Digital Accessibility Analyst at Salesforce's Office of Accessibility. With refreshing honesty, she shares how her own visual impairment and an unexpected quality assurance ticket requiring screen reader testing sparked her passion for digital accessibility.

What makes Crystal's work particularly groundbreaking is her exclusive focus on internal accessibility for Salesforce employees. While many companies prioritize customer-facing accessibility, Crystal champions a more holistic approach: "You can't really say you're accessible if your employees are having issues with accessibility trying to provide accessibility for your customers." Her role involves ensuring all internal digital tools, resources, and documentation are accessible to everyone, especially employees with disabilities.

Crystal offers a compelling vision for accessibility education that starts far earlier than most would consider. Rather than beginning at university or in professional settings, she advocates introducing accessibility concepts when children first encounter technology. By comparing accessibility features to video game "mods" when speaking to school children, she makes these concepts relatable and plants the seeds for a more inclusive technological future.

The conversation takes a fascinating turn when discussing artificial intelligence and accessibility. Crystal provides nuanced insights into how AI systems trained on existing internet data may perpetuate accessibility problems without expert oversight. She highlights how technical literacy gaps and science fiction expectations create challenges in critically evaluating AI's capabilities and limitations.

Whether you're an accessibility professional, educator, or technology enthusiast, this episode offers invaluable perspectives on creating truly inclusive digital experiences from the inside out. Subscribe now and join the conversation about building technology that works for everyone.

Chapters

• 0:00 Introduction to Crystal Preston-Watson
• 9:44 Journey into Accessibility

Transcript

Neil Milliken: 0:00

Hello and AXSChat to . I'm delighted that we're joined today by Crystal Preston-Watson, who is a Senior Digital Accessibility Analyst at Salesforce within the Office of Accessibility. I've actually been following Crystal for a long time on Twitter, now Blues ky, and had the pleasure of being a presenter on a series of presentations of 18 months where you were doing some stuff about vibe coding at the time and how accessible the different AI engines were at producing code and I thought, oh, this is someone that I really want to get on AXSChat . It's taken a little while, but I'm glad that, despite the best efforts and acts of God and maybe you'll explain that later but that you're here today with us, so welcome. So yeah, can you tell us a little bit about how you came to work in the field of accessibility?

Crystal Preston-Watson: 0:58

Yeah. So I started like let me kind of like because it's a long, windy story, but I'll try to condense it so I was a several time dropout of college and eventually I was going to school for journalism. I dropped out to actually go work in journalism for Alternative Weekly, where I am based right now, which is Denver, colorado, for Alternative Weekly, where I am based right now, which is Denver, colorado. And you know, this was also during the time right when kind of newspapers started to suffer from lack of readership and you know like you know kind of profiting and things like that. So I was in working as a web editor, mainly because on my free time I would code, because that was what I did. I taught myself how to code and do stuff like you know, like HTML, css, javascript. That's what I did instead of going to class. So I dropped out of my journalism program to go work as a web editor, mainly based on my coding skills. And then the you know kind of the the recession hit Sorry, I have a cat who's very demanding of attention. So I and I was like, okay, this is not sustainable. So I went to go work in tech as a front-end developer and then, after around 2013,. I decided I was like I don't know if I want to stay a front-end developer. So I did kind of the reverse, which a lot of people do. I ended up going from front-end development to quality engineering and really enjoyed that quality engineering and really enjoy that.

Crystal Preston-Watson: 2:44

And then, as you know, as I was going through and you know kind of you know through my quality engineering career, that's when I came upon my first like ticket, you know, accessibility ticket. I remember I was, I was, you know, marnell, get in, see what's like. You know what's been assigned to me. And it was says test with Jaws. And I was like awesome, what is Jaws? What am I going to do with that? And the funny thing is I have I, you know, you can see that I have a eye patch on and I've always had only really one functional eye because I had ocular toxoplasmosis. I was born with it and it, you know, affected my right eye. And so time where I got assigned this ticket, I was suffering from vision, more vision loss in my left eye and the ticket kind of really like just inspired me and I was like one. I was like I think I should look more into this because I might need this, given the things that are happening. Two, I was like, well, even though I'm going to look more into this, I cannot test this ticket, and so I sent the ticket back.

Crystal Preston-Watson: 3:56

But I realized I needed to learn more about accessibility. That's kind of where that started and that was around like 2015. And then from there I moved over to Aetna, like I, you know, before, I was joined with CVS and I was there as a, you know, mobile quality assurance engineer. There you know, this is before WCAG had the you know were the mobile success criteria, so I was trying to translate you know 2.0 into how that would translate into native mobile applications. And then, that's just like from there and you know, moved to a startup for a little bit and then ended up at Salesforce. So I know that was long.

Neil Milliken: 4:47

But we love to hear how people ended up in our industry because, let's face it, a lot of people don't start their careers thinking I want to be an accessibility specialist. I'm hoping that that's something that's coming along and we're working on that in various different spaces. But most people have had a fairly winding route into accessibility and there's usually some kind of personal or familial motivation for working in the field. So it's always good for us to find out how you arrived at the point that you're at now. So thank you for sharing your story with us, and I'm sure Debra's got a question, and yes, indeed, she does have a question. So I'll over to you, Debra.

Debra Ruh: 5:31

No, go ahead, Neil, I can wait.

Neil Milliken: 5:34

Okay, so I was just going to ask. We've had the backstory, so what are the kind of things that you work on now? You said you're working in the Office of Accessibility, so that's the sort of cao function within your organization yeah, so it's actually.

Crystal Preston-Watson: 5:51

This is so. You know, it's a lot different from what I did at um. It's partially different from what I did at edna, even when um I started at sealsports, because I started as a quality engineer working on a specific application and then after two years that's when I moved over into this position, and this position is a little bit unique. It's starting to be found in other kind of, especially bigger corporations and companies. I'm completely internally focused, so I'm tasked, you know tasked and responsible for internal digital accessibility for our employees. So this means any sort of digital tool, resource, documentation that you know employees need to do their job. You know we want to make sure that they're accessible to everyone, you know, you know, especially our employees with disabilities. So if there's that and it runs the gamut of one making, vetting and testing and reviewing applications before they are used by employees, and if there's issues after they're used by employees and there's accessibility issues, you know remediating those with. You know if it's internally built or if it's a third party, you know software tool, getting those talking to different people, and so, yeah, it's it really and also not just kind of you know the. You know, oh, I'm going to fix this or reviewing this, but also, partially, you know, helping teams to before they even start building out tools. Of learning about you know accessibility especially and this is different from you of learning about you know accessibility especially and this is different from you know product. You know because there's a product accessibility team which are great and they work on Salesforce products, but there's there's Salesforce is big and so there are a lot of different teams. You know that you know are don't work on our products specifically that build internal tools and also need to know about accessibility.

Crystal Preston-Watson: 7:47

So that is kind of my job and it's really it's allowed me to really expand and you know and grow a lot of my skills from just the technical skills and also bring in some of that those journalism skills that I had from you know previous, you know from my time working in newspapers, and even improv skills from you know because my hobby is doing improv comedy, so doing a lot of like talks and things like that, having especially as someone who has social anxiety disorder. So you know being able to bring those improv skills in to help me get through. Sometimes, you know, pretty scary to me, scary you know talks Even this is scary to me Like so. But yeah, that is, that is my job and I know there's starting to be more and I hope there is because you know we, you know accessibility is not just user, you know like user, customer focus. It needs to also be internally focused with employees, because you can't really say you're really accessible if your employees are having issues with accessibility. Trying to provide accessibility for your customers and users.

Debra Ruh: 8:58

Crystal, you have such an impressive story and, as Neil said, we love to hear how people got in this field, because people don't get in this field in traditional ways. Also, as a former programmer, I can't believe you taught yourself all those languages. I'm just impressed. Your brain must be so interesting that you could teach yourself that. Congratulations, Wow.

Crystal Preston-Watson: 9:21

I'm a little bit rusty. Now I'm actually starting to reteach myself because I'm more of a nuller JavaScript and I know enough of React to be very dangerous and I realize I'm like, oh, I get to get back in there because all this new stuff coming in and I don't get to code and do stuff as much as I used to.

Debra Ruh: 9:47

I coded in a language nobody's heard of, but it was tied to the mortgage banking. It was called EasyTree and every once in a while somebody will go oh yeah, I've heard it, but most of them never heard it. It was still programming language. But so I just wanted to comment on what you're doing internally, because I think that's so powerful, that Salesforce is making sure their employees are taken care of. And I will say you said I know there's more big brands doing this, but I will be honest with you, I don't see it. And maybe there are.

Debra Ruh: 10:18

I'm sure Atos is doing it. They're usually right on top of things, but I will tell you, crystal, I don't see that. I hope it's true. I hope it's true. I hope it's true because, as you said, I remember talking to a client one time and I'll even say the client, it was Capital One years ago, and they said we want to take care of our employees.

Debra Ruh: 10:35

We really do, but we have good relationship with our employees and so right now we're going to start at the time, making sure customers are taken care of, and I think a lot of people did that. But then they forgot to go back and take care of the employees, and I think Capital One was a really good example of taking care of everything. But I'm just saying it's important to take care of your customers, yes, but also taking care of your employees will allow you to take care of your customers better. So I am hoping that it's true, crystal, that you are seeing more people doing this. I'm just not seeing it. Maybe I'm looking in the wrong direction. So that was the question I had. So are you you do think this is starting to happen more, because I really wish it would happen more and I'm glad you're there.

Crystal Preston-Watson: 11:19

I think and you know that's a good thing I guess I get people asking me and it's more, they're doing it off the side of their desk, kind of like how traditionally accessibility for customers really know, like people will you know kind of mention to me and they're like, hey, I want to do, I want to do something like this. At our company and I was at I gave a talk at Xcon earlier this year that just explained about Salesforce, like the internal digital accessibility program, and one of my biggest reasons for doing that is to really, you know one, inspire people to start thinking about that and to give that blueprint of, like you know, when they do, you know, go to their you know stakeholders and C-level executives that they have something to point to, of saying, hey, this is being done, it's you know, it's growing, it's really, you know, something that is a good investment to have.

Antonio Santos: 12:27

So, crystal, you have been. I'm sure you talked with a lot of people. You know you mentioned the conferences and the keynotes that you were given, but I want and we know that over the last couple of months we have seen some disruption in the area where we work. I don't want to go anyone into that that can put anyone in trouble, but I would like to ask you in what areas we are not doing enough in order to make sure that we improve and the web really becomes accessible.

Crystal Preston-Watson: 12:57

I think the main thing is that we really do need to start getting people, when they first learn, first learn any sort of computer skills. And something that I've been kind of really really trying to push is not just like universities and college, when you know they're like computer science or like, oh, let's get accessibility in there. Or coding boot camps of like, oh, we'll do. You know a unit there, first grade, kindergarten, whenever a school, like whenever someone is officially doing you know that class accessibility is a part of it. Because I think you know you can only do so much retro like you know retroactively, once people get to the point where they're building you know products and services. But to really have it ingrained that accessibility is important, you have to start. You know young and you know. That's why it's great to see. You know brands like Roblox have like an accessibility page Also, you know. You know it's the reason why I'm really you know I really champion video game. You know companies and you know for systems and things like that, having really highlighting accessibility, because that is getting people where they're not thinking about that. That's not something and it's getting to think about it and it's getting them to realize how much accessibility is important and it really helps them for someone who may not know. You know, because a lot of times, like some people don't know they have a disability until one day they turn on a like a setting and they're like, whoa, this is, you know, has helped like videos, while now a lot of younger people have those like they will not watch anything without having captions.

Crystal Preston-Watson: 14:49

You know people who didn't realize they were colorblind when they're playing, like, say, video games, and they're like, you know, use. You know I've heard that where they end up with some games have like a setting for users are colorblind and they're like, oh, wow, this has improved the experience. And that's something I, when I've volunteered, like with you know, grade school kids, I will use the analogy of mods in games. I'm like I play a lot of games like Skyrim and things like that. So I always talk to them I'm like, yeah, accessibility is, they're like mods.

Crystal Preston-Watson: 15:24

You enjoy the game, but you know, sometimes you know mods really just you know you know one. Either you know elevate the game or you know if there was something not there in the base game can help Remedy, do that as well, and that really starts to click into their heads, and so that's something I really would like to see is just going beyond. You know people who are in, who are working age or in in companies or even college and universities and really get into K through 12, you know preschool again, whenever, whenever the you know that someone first picks up a, a smartphone.

Neil Milliken: 16:05

You know a laptop or whatever the sort of learning about mods and accessibility features, and just the fact that people use things in different ways and that we need to make our tech work for those different ways, then, is setting us up for future success as a society, because this is going to be something that people understand needs to be done by default, whereas I think that we are living with the impact of decades or centuries of previously inbuilt ableism and everything else which was ingrained in the attitudes and ingrained in the way that we do things and the way that we do business, and it takes time to turn all of that around. Inclusion is not just a whole society effort, it's an intergenerational effort as well, and we're kidding ourselves if we think we can fix it overnight.

Crystal Preston-Watson: 17:19

Yeah, I mean, it would be wonderful. Fortunately we live in a reality.

Neil Milliken: 17:23

Well, maybe we could just ask AI to do it, and I think maybe that could take us to the next sort of thing I'd love to talk to you about. As I said, I saw you playing with AI. Obviously, you've been curious about the impact of AI on coding and so on. Wasn't that great the last time you were demonstrating it? And now we see lots of memes on various social media channels about vibe coding and all the rest of it. Do you think that large language models are getting better at producing accessible code and if not, what do you think needs to happen for them to be producing something that is more inclusive?

Crystal Preston-Watson: 18:10

it's funny because I didn't you know back when I did that talk it really you know vibe coding. It wasn't called vibe, like vibe, you know, and I you know, and I had only just learned of this like right, when this term started to become popular. You know, it's like I think it's just so funny. It's like you know no coding, just vibes and nuts it's. You know this is such a really you know it's such a complicated thing of when you're talking, especially when you're like AI and accessibility, you know and it's you know cause you have, you have your the defenders, you have the you know detractors. I think you know, and something that I've, you know, I have definitely talked about and I you know in different talks and and you know, in like my own, like writing, like my own posts on my site, is that I really the one thing is is that AI, in certain forms, I think can be beneficial for accessibility, but I don't think it's one it's not going to solve, it is not the overnight solution. I think one, especially when it comes to coding, is that where's this data coming from? Is this data correct? And I think by the very fact that the Internet, even you know, really before this. You know AI kind of really came to this. You know, really, you know forefront was not very accessible and there wasn't. You know there's a lot of, you know, misconceptions, there's a lot of ableism and that stuff gets pulled into data sets and large language models and you know it can. And so if you're letting certain you know models like just go wild, you know AI going wild and they have no knowledge about accessibility, they go to one of these watch language models and say, hey, make me an accessible form. That form it might put something out, and they're like, okay, this is accessible. But you know, if you're someone like you know any one of us or you know, and you look at that same form, you could probably see many issues.

Crystal Preston-Watson: 20:35

And that's something that I did show, like the prompts, because it's like, well, you know, because it's all about, oh, you have to have good prompts to make sure you get really, you know, get what you want. Well, how can you get what you, you know get something really good when you don't have that background knowledge of accessibility? And so, and then you're trusting this, you know this model to be, and it's like, but who knows, like is the? You know, is the?

Crystal Preston-Watson: 21:04

Is these data states, data kind of you know, are they being, I don't want to say judged, but are they, you know, really, you know the data is not tainted, or you know, or dirty or anything like that? Um, because that's the thing is that when you don't have that, that knowledge, then things like that, can you know, slip in and people who are trying their best and who want to, you know, to their, you know, they want to be accessible, they, they care about accessibility, they're thinking they're doing good and they're just perpetuating, you know, kind of like what you know, the status quo, that is not, that is not accessible, but they don't know that so as a quick follow-up on that.

Neil Milliken: 21:50

So I understand that they, you know not.

Neil Milliken: 21:53

Not everybody knows how to write prompts yet, and not certainly not everybody knows how to write prompts around, how to tell a large language model what it needs to do in order to be successful, telling it to act as an accessibility developer, telling it to reference the relevant accessibility code examples and the repos and follow guidelines and all of this kind of stuff.

Neil Milliken: 22:16

So you're asking people. The expectation is for people to follow and create pretty complex prompts, which goes against how most people expect to be using AI at the moment, right? So people that are deep into it write complex prompts and understand that you have to do that in order. So prompting is just another type of coding. It's just using natural language rather than programming language, and I think that people don't really understand that yet. I just wonder whether or not actually, because the language models also have underlying prompts that are in build, whether we think it's the responsibility of the people that provide those large language models to kind of build that in and say, oh, you want a website, well, therefore, you need to follow these inbuilt requirements that tell you to do that, so that we're not putting the oomph on the end user.

Crystal Preston-Watson: 23:08

Yeah, I think now really a lot is on the user models. You know something and then they'll just post it directly and you know, and you realize that, and they're like okay, here this is correct, and they take the, the model at word, as as really like law and you know, and they, you know, and they don't understand because this is something they're like, they, they assume it's, that's that assumption that has always been there, that tech and computers are correct, they're infallible, and it's like that's not true, because who creates the tech? That's humans. So tech is not infallible, it's not error-free, and you know a lot of we.

Crystal Preston-Watson: 24:34

Really there is a lack like, oh, I want to, um, create a, create a recipe for me, like they're not looking up, of how to write a prompt, and then that's when you get something where it says, oh yeah, boil a bunch of rock, and that's like and because you know they don't know how to write it, so it excludes certain things, or things like that the onus is on, you know, on the, on the, on the user, and the thing is is that you know, if you want that to be on a user, then you have to provide that education, otherwise then you kind of need to be in charge of that.

Crystal Preston-Watson: 25:15

So it's like you can't. It's like you can't have it, like you know you can't have it. Where you don't provide that, you know that that guidance, if you don't provide that education, provide that education. So it's like, if you want things you know, be able to do this correctly, then teach them how to do prompts in a language you know, in language, plain language, that they can understand how to do that. Or if you don't, if you're not going to do that, then it goes back on to the provider of these you know LLMs.

Neil Milliken: 25:49

And Tony, did you have a follow up?

Antonio Santos: 25:54

No, yeah, I was just looking at the time just to make sure that we were really on it. I think, in the area that you're talking, I think we need a lot of critical thinking in order to evaluate. But, however, every day we see a founder or someone with lots of responsibility for language models saying all types of hypes that are not, that really are between true and not as accurate. So people just take their word as a kind of a ultimate true. They say okay, if they are saying this, I'm sure I can trust. The language model is the latest tech, so I don't care what the experts are saying, I just the guru knows it all. So I think that's a risk that we face today is that people just believe because they want to believe and there's not enough critical thinking when we look at the outcomes and when we work with large language models, whatever the model we're talking about here.

Crystal Preston-Watson: 26:49

Yeah, and something of this actually goes into media literacy, because they're, you know, the sci, like you know, sci-fi movies, movies, you know. Like you know that have, like you know, blade Runner, like you know, have cyberpunk elements and things like that. People really they're excited about that. They love those type of movies and they want that. They want love those type of and they want that, they want to live in that type of reality. And so, you know, when things are promised that says that is here now, people want to believe that and not, you know, and so they may not question of, like you know, it's like is this really, you know, is this really possible? This is like because it's like, well, I don't know, you know, because people tend to downgrade their you know what they know, if they feel, especially if something is really technical. You know where it's like, well, I, you know how could I? You know, I'm just this person, I don't understand things. This person is so much smarter than me, and so they're like, they're saying the future is now and so I'm going to just put my, my faith and trust into that, you know, into that, and you know it's yeah, and so that comes with like the wishful thinking of, of wanting, you know, yeah, the future to be here and, and because that what's promised in like in, you know, in in fiction and you know in stories and stuff like that that have this, you know where tech is just like super cool and makes things a lot easier. They want that and I understand that too. There's a lot.

Crystal Preston-Watson: 28:18

I used to watch the cartoons like was it Chuck Jones that used to do like Looney Tunes, and they have like all the future with all the you know they had like the refrigerator, like popping out different things, and like you know it was a promise that you know you didn't have the robots do everything. And I'm like I remember watching it as a kid. I'm like, oh man, that's going to be super cool, and I realized sometimes they would say 2025. And I'm like now I'm thinking that didn't happen. There's not a food pill that's going to pop in my mouth and things like that. And you know as. But just think if I was, you know, as a kid, someone's like, hey, this is now. This stuff is going to be. You know, you don't have to wait to 2025. It's going to be here right now and I think that's what's happening.

Crystal Preston-Watson: 29:02

Is that you, the emergence of the lack of technical like literacy from you know, and also the lack of media like literacy are merging, so you're getting this kind of really big ball of like, complex ball of like what is going on and what can be happen, and that I realized we just say AI. A lot of people don't realize that you know there's the, you know there's generative AI, there's this, you know discriminative AI and you know there's AGI, agi, addict and all these different things, and so the fact that you know it's like well, discriminative AI, you've already been using that. You know that's like. You know things that happen for like spam filters and things like that. So when we're like all AI is, you know all AI is bad, it's like no, it's not, because you know AI is not this thing. That just happened three, you know three. You know two to three years ago it's been you know. You know. You know it's been around. You know a lot of before it was really just called AI. It was you know. You know machine learning. I mean it's part of it's all you know and that's again that comes on.

Crystal Preston-Watson: 30:16

The whole thing of like not having tech literacy and being able to understand that, and I think there is a responsibility for those who us that are in tech, to really kind of help spread that knowledge and not try to gatekeep it and not try to be like, well, you're not going to understand that because you know, you don't, and it's like I don't. Like that, I hate. I hate that type of thinking of like you know, so it's like I, you know, and that's why I try is you know, you know, when I like, when people ask me that you know, you know, friends, moms are like you know, explain to me, then I will go ahead and do that, because I, you know, I'm not going to. You know, I'm not going to sit here and assume that they're not going to understand this. It's like that's not true. You just need to explain it in a way they're going to understand, like anything, yeah.

Neil Milliken: 31:06

And it's important that we do.

Debra Ruh: 31:08

Still, I see you have Lieutenant Hurra behind you, but she was my favorite favorite on Star Trek, and we also didn't ever get around to beaming me up, scotty either, while we're doing this technology. So I just made some brilliant points, because people are so afraid of AI when the reality is, you know, we've already been doing this. This is like machine learning. We're just presenting it to you in a different way but at the same time, not saying we're not walking some really scary things right now. But I just want to make one more quick comment.

Debra Ruh: 31:40

I know we went over, but as a technologist, I've also been a technologist my whole career and the reality is we, as technologists, we need to care enough about the humans to make sure the humans are fully included, and I'm seeing that not happen in ways that troubles me. So I'm really, really glad that we have leaders like you out there, because it really is about human inclusion in every single phase. What does human inclusion look like? Times I might be over a certain age, sometimes I might not be able to see in the same way. It's about humans. So thank you, crystal. Over to you, neil.

Neil Milliken: 32:20

Absolutely Fully agree. Thank you so much for coming on. It's been really great to be able to chat with you and share what you're doing, and I'm sure we'd love to have you back at some point as well. So thank you so much, crystal. It's been a real pleasure, and I also need to thank our friends at Amazon for supporting us to stay on air all of this time.