How Amazon Builds Workplace Accessibility For 1.5 Million Employees

Accessibility at work is easy to talk about and hard to deliver, especially when your “office” includes fulfillment centers, delivery stations, corporate teams, and cutting-edge tech roles spread across the globe. We sit down with Megan Smith, Global Accommodations and People Accessibility Lead within Amazonian Experiences and Technology, to unpack what it really takes to run accommodations and workplace accessibility programs at serious scale and with real speed. Megan shares her own path into disability inclusion as a legally blind leader, and why self-advocacy plus operational rigor is a powerful combination.

We walk through the full accommodations flywheel: intake, evaluation, decisioning with managers and sites, and making sure the accommodation is implemented. Megan explains why much of the volume comes from temporary injuries that need fast, consistent handling, how pregnancy-related accommodations require dedicated expertise, and why complex disabilities benefit from high-touch support that adapts to different job environments. We also explore how people accessibility focuses on preventing barriers in the first place by setting standards, offering tooling, and influencing what the company builds and buys for employees.

Then we look forward. AI is already blurring the line between assistive technology and personalization, with big implications for screen readers, captions, and the idea of a “one console” work experience. Megan breaks down what must remain true today (solid accessibility fundamentals and WCAG principles) and what might change tomorrow, including trust, privacy, and how we filter signal from AI-generated noise. If you care about disability inclusion, HR accommodations, accessible technology, or the future of AI accessibility, this conversation gives you practical mental models and sharp questions to take back to your team. 
Subscribe, share this with a colleague, and leave a review so more people can find the show.

Transcript

Welcome And Megan’s Path In

Neil Milliken 0:00

Hello and welcome to AXSChat. I'm delighted that we're joined today by Megan Smith. Megan is the global accommodations and people accessibility lead within Amazonian experiences and technology. Now that's quite a long job title. Amazon is an enormous employer, it has about 1.5 million niche employees around the globe. So this is a big chunky role. So Megan, welcome to Access Chat. It's great to have you on. Can you tell us a little bit about yourself and about how you came to be working in the accessibility space? And then we'll go on to talk about what you do in your role as the accommodations lead within Amazon.

Megan Smith 0:43

Happy too. Thank you so much for having me. I'm excited to be here and chat with you. A little bit about me. So I've been with Amazon almost eight years, almost double digits. And I joined to start what has become people accessibility, which is a part of my scope still today. I got into the field of disability and inclusion and accessibility somewhat on accident. I'm legally blind. I have a progressive retinal disease. And I got my first guide dog when I was 17. And that is a very character-building experience and age to get a guide dog. I learned. And so I was kind of thrust into self-advocacy and kind of just became really interested and passionate about, you know, ensuring that people had what they needed, that we were thinking through kind of where disparate impact or where adverse treatment of people with disabilities was kind of evident around me. I was working in healthcare at the time, and it was not lost on me that kind of everyone is at some sort of disadvantage in a hospital. And then when you have a disability, that can can become compounding. And so got really interested then. And it's a super long story, but ultimately recognized that I didn't want to just have my lived experience to stand on, but I wanted to understand kind of disability more broadly. And so I got a graduate degree in disability services and ultimately began working for a health insurance company in kind of the broader diversity space. The very first thing that I did there was get laid off and wrote a note to the CEO that basically said, you know, I understand your mental model around diversity. His point back then was it shouldn't be one team's job, it should be everyone's job. And so I said, I hear you, but you know, you just hired me as a diversity analyst. Here's what I'm seeing. And through that conversation, he said, you know, I'm not sure what role you're going to have, but you will be here doing important work and influencing my leadership team. And so I was able to work with the C-suite for several years at that company. And often, as my role progressed, I moved into broader customer experience. And often as we were talking about customer experience and what we should aspire to, companies like Amazon were heavily referenced. And I was actually well connected with Disability In and through them kind of learned that Amazon was hiring for this role and it happened. So that's how I got here.

What People Accessibility Means At Amazon

Neil Milliken 3:05

Fantastic. And I know that that Amazon is customer obsessed, but the employees are the internal customers, and that means you've got a pretty large customer base to serve across all different parts of the globe. So what are some of the things that you do in your role to support Amazonians in the wide variety of different roles? Because Amazon is enormous. You've got so many different types of companies, so many technologies, you've got a range of things from your fulfillment centers to your logistics network to people that are coding and planning and wanting to design drone-based deliveries and satellite networks. So there's a huge diversity of people and roles and so on. So how do you then deliver disability inclusion to that wide client base of internal customers?

Megan Smith 4:02

Yeah, I would say the breadth of roles that we have and kind of how we build things make this role really, really interesting. And so just to kind of sort through what's in my scope, I'm accountable for accommodations case management. So as soon as you recognize that you need an adjustment to when, how, or where you work, my team handles kind of the discovery of that, the intake, the evaluation. We work with your site or your manager to get to a decision, and then we help ensure that that accommodation is implemented. It's kind of that end-to-end accommodation process. And we do that for medical, religious, and statutory requirements for flexibility. And so it extends a bit beyond medical. Then we are accountable for direct delivery of accommodation services. And so you mentioned, you know, 1.x million employees. We have a lot of employees who are blind or who are deaf or who are neurodiverse and rely on human services, on assistive technology. And so my team is accountable for delivering those services, whether that's an interpreter, whether that's a screen reader. We make sure that we have the services and that we're deploying them at scale for Amazon. So that's another part of my organization. And then I kind of close out that flywheel by being accountable for people accessibility. So my team is, we don't write the code for products that Amazonians use, but we set the standard and offer services and tooling to ensure that what we build and what we buy for Amazonians is accessible. And so I think about it as people accessibility, our job is to help prevent a barrier from existing in the first place. And an accommodation is if a barrier does exist, we remove that barrier for the employee. We learn from those accommodations and that informs how what we focus on in accessibility. And so not directly accountable for disability inclusion, but certainly a part of that broader landscape at Amazon that's working to ensure that people with disabilities are able to deliver and delight our customers in a lot of the many ways that you described.

Debra Ruh 6:04

Megan, well, welcome to the show. As Neil said, I do understand that Amazon is employ customer obsessed. And I don't know what I would have done. I don't know what we'd have done here in the United States without Amazon during COVID because y'all were such a blessing to the community during that time. I know that a lot of the employees are in the United States, but you still have a ton of employees outside, all over the world. And I know some people call accommodations adaptions in other, you know, areas, but it just feels like I'm curious about things like I know that in the past, people to get an accommodation, they needed a doctor's note and things like that. And I just in my mind, I think you have such a big important job. And I just am wondering how do you how how could you possibly get your hands around such a big job? I'm very grateful that you're doing it. I'm also really grateful for what you said about the CEO before, because most of us will be like, fine, we'll just leave. But you took the time to say, wow, you were missing such an opportunity here. So I wanted to applaud you for that bravery too. But you have a really, really big job. And I was just wondering how that works.

Megan Smith 7:17

Yeah, thank you. He described me as having chutzpah, and it's something I've tried to carry, you know, these last 20 years with me as like a core trait that I make sure I don't compromise much on. Yeah, it's, you know, the breadth that Neil mentioned and then the scale that you mentioned, right? And kind of the cultural norms, the systemic opportunities or lack thereof, it all interplays with how we deliver accommodations. Accessibility, I think we benefit from the fact that a lot of what we do at Amazon is global, kind of by design. And so that's a bit easier to handle. But accommodations, they can be very local. And so we have a team of not going to quote numbers, but we have a large team and we build technology and products that help support us delivering these kinds of accommodations at scale. The team has undergone an incredible transformation lately. They started

Building Accommodations For Speed And Scale

Megan Smith 8:10

it a year and a half ago, and it's just proven to be a really strong model for the way that we work. I mentioned we do kind of a broad swath of accommodations. So if you need an adjustment, we are likely where you go, Med Amazon. And so a large portion of our volume is people who got hurt over the weekend or who've had that nagging back pain forever. And right, it's a temporary, maybe musculoskeletal, and they just need some time to be able to work a little bit differently until they're they're 100%. And so this kind of temporary category is actually a lot of what we do. And those accommodations need to happen fast. They, you know, they have a lot in common in terms of how we work through them. And so we have a really tight kind of milestone model that we run those through where we've got experts that kind of understand each step of the process and get employees from I need help to I'm working with my accommodations really quickly. Then another big bucket that we cover is pregnancy-related accommodations. And so we ensure that we understand the rights of pregnant workers, the experience in this kind of critical moment in an employee's life. And we make sure that we're delivering accommodations for them, whether that's additional breaks, maybe for lactation, or maybe you know, lifting restrictions, we make sure that we're delivering that. And so that kind of comes through its only separate path. And then we have employees who have more complex disabilities, ones like mine, right? And so for that, we have a high-touch accommodation team. And so these this is where we've really developed keen expertise and what it's like not just to kind of have these disabilities and common accommodations, but what does that look like in an AR sortable building or in a corporate building or in a delivery station? And so the team really works to understand the ways in which we work at Amazon and how that shows up with various disabilities and curates accommodations for that. And then they're they're there in that kind of relationship with the employee to help them as their needs may change. That's an area I want to continue to to to improve upon, right? Like we're all kind of in this continuum of disability and our ability to support people as their disability evolves, or if you've never had a disability and it's new to you, right? And so that's really kind of an a next stage in our journey that I'm excited to unlock is kind of how we maintain that relationship with with our employees.

Debra Ruh 10:33

Megan, I I know that you that you brought up so many powerful points. And I just want to say that one thing that when I was uh learning all of this, one thing that I was talking about a lot would be the temporary ones because people didn't really understand it. They would get afraid when I was talking about maybe somebody with a permanent disability, but I think people understood that temporary a little bit better. So that is a I just wanted to applaud your answer. I'm I'm very impressed. Over to Antonia.

Antonio Vieira Santos 11:04

So welcome, welcome, Megan. So you are working in this space for for quite some time. You work in in you know in different companies. Do you feel today that you you still need to explain the true value of disability and inclusion? If yes, how do you get the buy-in from leaders that might not understand it?

Megan Smith 11:24

Yeah, I love this question. And anyone that that I've ever interviewed for Amazon for a role related to this has heard this from me. Not to say anything negative or even neutral about any previous company. What I will say is true about Amazon, though, is that this is not a place where I feel like I need to defend the why. You mentioned customer obsession, that's one of our leadership principles. We have many others, and so many of them, probably all of them, if I really like took time to go through them, map back to the business case for disability. So we have one, for example, insist on the highest standards. Something I'll pull out of that is that we never send a defect down the line. And for anyone who's talked about shift left and accessibility, that is the definition of insist on higher highest standards. And so I need accessibility so that I can avoid the need for an accommodation that's working around a barrier that should have never been there in the first place. Right. So that's just one example. Customer obsession is another easy, obvious one. We cannot deliver in the way that we do for our customers without a workforce that understands, is committed to working backwards from those customers and that holds a high bar. And so I need to be able to ensure that those Amazonians who are bar raising by definition of our hiring practice can actually do their jobs. It's my job to get all of those barriers out of the way so that we can unlock them to deliver for our customers. I'll say another invent and simplify. This has been a really fun conversation as we've been talking about artificial intelligence across all of the ways that Amazon may use it. And there are so many proof points where individuals with disabilities, the one I've been talking about lately, is screen reader users, right? And so we see kind of this shift for people who are predominantly blind low vision using screen readers, where as this technology emerged, adoption was really low. It hovered around, you know, single digits and then started to creep into double digits. Once it started to hit close to 20%, it skyrocketed. And that was a new kind of interaction model. And what I've learned is that as this adoption was happening, studies would show that kind of blind screen reader users actually had changes in their neural networks, changes in the way they processed speech. And so for those of you that have engaged with someone who's a screen reader user, they're listening to that screen reader two, three, four X, what someone who is not a screen reader user is. Those of you that might listen to a podcast, you're probably not listening at 1x, or you might not be listening at 1x. It might be 1.25, 1.5, maybe 2.x if you're like an active consumer of audio. But what does that tell us about kind of the way that we're all engaging with technology two years from now, five years from now? And how much will that personalization matter? And I think that in this example, blind people who use screen readers are a good leading indicator of what we can expect. Captions are another example. And so I think we just have all of these opportunities to learn from people who have been inventing their way around barriers for a world that wasn't designed to that for them. And instead of pointing that learning backwards, oh, what would we do different for disabled people? We can point it forward and say, what have disabled people taught us that will matter to everyone going forward? And so that's another great example of invent and simplify. Again, I can go through every leadership principle. And so here it's not why should we? It's how do we? How do we do it at our breadth? How do we do it at our scale? How do we do it in a way that meaningfully uh improves the lives of Amazonians? And so, Antonio, to answer your question, it's not, you know, it's not part of the story here.

Winning Buy-In Without Selling The Why

Megan Smith 15:00

And I'm really grateful for that because then I get to spend my team gets to spend our time on actually doing the work instead of convincing leaders that the work is important.

Debra Ruh 15:09

Megan, I'm really impressed because to be honest, I think our community doesn't understand the kind of depth that's happening in Amazon. So, first of all, I'm very impressed with what I'm hearing. Because I think a lot of us feel that so much, we've lost so much during the attack on the DEI that yeah, somebody started. But uh I I just I'm very impressed with what you're saying because it it's it's the only way forward. So, first of all, thank y'all for the leadership. I know that obviously Amazon's very involved in AI, and I've I'm a little tiny point in that I'm one of your beta testers and I'm always giving my Alexa a really hard time. I shouldn't say that word, but it but it's also fun training it and learning it and watching y'all grow. It's just very interesting times. But I'm also hopeful that AI has the ability to really reduce some of the gaps that we see for people with disabilities. And so I was curious, you know, how is AI helping confusing? I mean, I know it's changing everything. So it's a really big question, but I was just curious, is it helping right now what y'all are doing?

Megan Smith 16:17

Yeah, I think it's it's a really interesting question and one that doesn't have an answer just yet, Deborah, uh, except the foregone conclusion that like, yes, with a ton of great like guardrails and and you know effort, right? It doesn't happen on accident. And we you I'm sure you all do mini podcasts on that. But I would say I was even just reflecting this morning before joining you all, my from 5 a.m. to 10 a.m. I'm you know working with my AI agent, right? I have you're the first humans I've talked to today in five hours, which is a little bit jarring if I'm honest. But I was reflecting on how kind of the speed with which I can work, the speed with which I can glean insights or make connections and data that I was never able to make before. So in those kind of basic and obvious ways that I think everyone is using AI now, absolutely. I think in my own personal use, right, my AI understands that I'm low vision, it understands my cognitive preferences, it understands when I'm getting overloaded or when I'm lost on a tangent, right? And so it begins to learn things that really blend the line between accessibility and personalization in a way that I think will open up access and accessibility to people who never knew that they needed it, right? Um and then I think it it I'm just interested to see where that goes, right? And again, all kinds of considerations like privacy and trust that are incredibly important. And so it's not a a tomorrow thing, but even in our own customizations and own use of AI, I think this is already happening just because we're human and we're interacting with something that by design wants to learn about us and help us. And so I'm seeing that. And then I think that there's, you know, there's a really interesting it I would say for us, and in just kind of one example that's coming to mind, there's the ability to get information, a kind of in accurate information in real time in a way that's relevant to folks. And so if you're really engaging with AI in a work context as your teammate, for example, and you're struggling with something, you know, maybe, maybe you do have one of those temporary injuries and you're needing something like, oh, I can't type. I could type on Friday. I broke my arm over the weekend, I can't type today. What are my options, right? You can engage with AI as long as it's got the right source of truth, as long as it's got the right guardrails, right? And you can learn things that are just, you don't know the word accommodation or adjustment. You don't know the word disability. You don't, that world is not familiar, you don't know assistive technology, but you suddenly have a tool where before maybe you had to go on a leave because you just didn't know and you didn't ask the right set of questions to the right person. And so that's where I think, you know, just from kind of how, again, we're we're blurring lines, like I think it will take away kind of some of the burden of needing to know what you don't know in order to benefit from accessibility, assistive technology, and accommodations.

Neil Milliken 19:11

That's super interesting. And and I know we've had some conversations prior to you coming on on about where AI is going, the blurring of the lines with assistive tech and and so on. And you you earlier said you know, that that you want to be solving the the problems of the future, looking forwards in instead of looking back. I think you know you're spending a lot of time in the console. So where do you think maybe looking at screen readers as an example rather than some of the other assistive technologies? Where do where do you think that that's going to develop? Do you think that people will spend uh less time using the screen reader or they're just sort of solely focusing on using the screen reader with the AI, and the AI is doing a lot of the other interacts with a lot of the other applications because you can use your AI to interface with lots of different uh other technologies. So you're in the workplace, you're using your AI, you can stay in the AI, you don't have to go into a spreadsheet, or you don't have to go into the Word document. You you know, you don't have to go to your emails and so on. Whereas before you'd be using your your assistive tech and switching between all of those things. Do you think that that that that's going to change A the way that people are using things? Because you're already doing that to a certain extent, but also change how people perceive AIs and when assistive tech and whether or not some disabilities will require specialist assistive techs in the future.

Megan Smith 20:43

Yeah, I've enjoyed these conversations with you when we've had them. I think I I don't know that any of us hold a crystal ball right now. I I think I I have a strong point of view on this. What I don't have a strong point of view on is the time horizon. And so I think it's important to say first that like the way that we think about accessibility today must still exist for the foreseeable future, right? It matters when someone engages with a device or with a website or a mobile app or your learning content or your email, that it works for them and their assistive technology. That matters a lot because we are not in a one console world yet. I think we'll get there one day, or at least the people who want to be there can be there one day. I don't know how long that can that horizon is, and I don't know if it's from zero to a hundred or zero to 70, but I do know that it's not tomorrow. And so the way that we think about accessibility today has to still exist. I also know that accessible code leads to better

AI Personalization Meets Assistive Technology

Megan Smith 21:40

AI outcomes, right? And so the things that will hold true, whether it's a human consuming them, right? I'm gonna get the poor from WCAG wrong, but perceiverable or I'm gonna just mess up all my words. The P O U R perceiver I can't.

Neil Milliken 21:55

I'll I'll do it for you. Perceivable, operable, understandable, robust.

Megan Smith 21:59

Thank you. Those matter to a human engaging with assistive technology. They also matter to your AI, right? Like your AI is going to be able to extract richer insights, more accurate information. And so I think it remains to be seen. Like, is 2.3.1.2.2 still incredibly powerful? Maybe. But are the overarching arching principles still relevant? Yes. I I think so. To your point of, you know, in we if we live in a world in a console, what matters most? And I think there it will be interesting what the community thinks about personalization and how much your AI is picking up on these things about you and what inferences it's making. And I think that's a to be determined. But, you know, I land on the side of like, learn everything you want about me, please, like be me. Right. Like I want a digital twin. I want to like really lean in where other people don't have that same kind of trust and openness or or desire. And so I think there's going to be a long time where it's kind of a customization of how much do I want to dial up what AIs can learn about me, therefore adapt to me. How much do I not want to? And I want to still use these kind of static, local, highly secure, highly private interactions. And I think it's to be seen. Your point, I think your last point around it it gets into, I think, again, something that you might have a ton of podcast content on around just how what will the nature of disability, the nature of assistive technology, is it sitting on my device or is it implanted in my brain? Is it right? Like I think that is just a world that is I'm excited to have those conversations. And I am super hopeful that forums like this continue those conversations and push those conversations.

Antonio Vieira Santos 23:47

I would say over the over the last 18 months, there have been at least two books about uh the quality of the web. And now the quality of the web has been going down as it used to be. So things are getting worse, more confused. Now with AI we we see slope everywhere. My question is when you want to express yourself as an individual outside work using technology that's available for you, do you feel that uh that you can do it? How do you feel that you can do it? Do you feel comfortable in doing with the tools that we have, or you feel more barriers, uh more stress, more anxiety in order to be able to have a voice in in the in the current times that we are facing now? I sorry, Antonio, I missed the like first core concept that you So there's two books published in the last 18 months that say that the web is getting worse. So when you want to express yourself in social networks or blogging, do you feel that you are in a good place or you are struggling more in order to be able to participate like you did five years ago for for a say?

Megan Smith 25:04

That's an interesting question. I'll you'll have to send me those books offline. I think for me personally, kind of I I think for me personally, the way I like if I think of a one-way engagement, me out into the world, I feel like the tools around me help, right? And so a couple of examples. I might have this is gonna be a very basic personal example, but I might have a photo of my daughter that I want to post. And I love Taylor Swift. And so I want to find the best Taylor Swift lyric to exemplify my daughter dancing in her first dance recital, right? And so like I can use AI to like more authentically be me in less time and with less cognitive effort, right? So super simple thing. I can also use Be My Eyes AI to generate an audio description of that image so that I'm posting that for my friends who uh use screen readers. And so, in those ways, right, like my authentic personal lived experience, kind of one way gets to be more accessible and more authentic and faster. And so I think in that way, that's working. I would say I think you made the point around AI slap and just kind of this notion of what's real and what isn't is becoming more elusive. And so I again I love to read those books. It is a challenging thing. Um, and even one of the things I'm learning just kind of in my personal use with AI is, you know, I think there's the change curve is really interesting. There's a period of time where you're like, oh, awesome, it took all my words and made them sound better. Then there's a period of time where you say, okay, let me give it less and let it extrapolate, right? That's kind of the productivity hack. Like there's a maybe just time hack, and then there's a like, I didn't have to write all those words in the first place. Now I can write less and it does more. And and then you're

AI Slop Signal And Trust Online

Megan Smith 26:55

in this like blissful area of like, oh, AI is amazing. And then you really inspect, right? Like, wait, where did that sentence come from? Or hey, someone asserted that was true, but it wasn't true, or it that's a different context, or or or. And so the way I engage with AI now, kind of in my, you know, day-to-day, whether it's Chat GPT, as I'm like trying to diagnose whatever skin rash my daughter might have, or, you know, in kind of researching topics for work or whatever it might be, I I kind of go back to this teammate concept of like, I'm not just I'm not outsourcing to it. I'm engaging with it, I'm teaching it, I'm questioning it. I'm again able to look all the way around a problem that I could only see in 2D before, but I'm not outsourcing the the judgment and the thinking in a way that I probably did for a week or two when I was like, this is amazing, right? Thankfully that was some time ago. But it's just this really, I think, interesting way to again look around a problem that that has felt unattainable or inaccessible before. And I I think that's kind of taking that to the meta level of like if we can kind of collectively get there so that we're not publishing as much slop because it is everywhere, and able to curate our context and our bar and our guide guardrails a bit more, then I think it's a like a more informed on both like slot might come out, but it can't come in, right? And so I'm interested in kind of that, right? Like, how do I put up my own controls while also hoping and trusting that others are putting up their controls? But but it's you know, I can I have more accountability and autonomy in what comes to me. I I think that direction is possible.

Neil Milliken 28:41

That's that's super interesting. Finding the ways to sift the the signal from the noise is is ever more important in a in a world where there's ever more information and working out ways to verify it. So, Megan, we've reached the end of our allotted time. It flies by when you've got three people asking you questions. Thank you so much for being our guest today. Thanks to Amazon for helping support us and keeping us on air. And we really look forward to sharing uh this episode with our audience. If you're a member of our audience and you like this show, please s think about subscribing to the podcast because everyone that subscribes helps get this message further.

Megan Smith 29:22

Thanks everyone.