Episode Summary

In this episode, Justin and Mark explore the future of accessibility with AI tools that are reshaping inclusive design and development. This episode covers:

• Building a general-purpose accessibility agent—and what we learned in the process – Eric Bailey
  • GitHub’s Accessibility Agent: Revolutionizing code reviews by integrating accessibility from the start.
  • Balancing Automation and Human Oversight:  Why automated tests aren’t enough and the role of strategic human review.
  • AI Deployment: Insights into using AI for accessibility testing and the importance of an orchestrator model.
• Anna E. Cook’s Workbook: Structural Workbook for AI + Accessible Design Systems
  • a focus on governance and bias mitigation.
• Accessible medical diagnostic equipment rule to take effect this summer – David Horowitz, MD.. FAAP
  • Accessible Medical Equipment: New HHS rules and the importance of consistent accessibility in healthcare technology.

Go to all episodes

Transcript

Accessibility Breakdown

Mark Miller: Hey, welcome to the Accessibility Breakdown. I am Mark, and this is

Justin Stockton: And I’m Justin.

Mark Miller: Are you intent? You’re just intentional. I always do, and this is. I have a script up here I’m reading. All right. So I am Mark, and this is good. Every week we pay tribute to those who stand out to us as leading by accessibility by picking three topics that strike us in some way. This week we’re gonna talk about building a general-purpose accessibility agent.

Justin Stockton: You never do this is. Last time.

Mark Miller: And what we learned in the process, this is by Eric Bailey, and this is essentially a new accessibility feature, I guess we could say in GitHub.

Justin Stockton: Yeah, yeah. New offering.

Mark Miller: Yeah. Okay. All right. and then we’re gonna talk about a structural workbook for AI, accessible design systems. This is by Anne E. Cook. This is interesting too. This is sort of essentially a playbook for understanding what your organization needs to consider before they start to use AI for coding.

Justin Stockton: and design systems.

Mark Miller: And design systems. And then we’ve got kind of a different topic here, and it’s accessible medical diagnostic equipment rule to take effect this summer. And this is very much related to Title II and HHS, but it applies a little bit differently than the sort of digital accessibility that we’re talking about, but there’s some crossover. There’s some interesting discussion to have there. So having said all that, Justin, I’m gonna count on you because a couple of these, these first couple of articles are pretty technical. And I’m gonna need you to really break it out. So let’s start with building general-purpose accessibility agent and what we learn in the process. I think this is really cool. I think this is a huge step up in accessibility for GitHub and it puts some really incredible things at people’s fingertips. So tell us what this is.

Justin Stockton: Yeah, so this is so Eric, what Eric wrote about here was basically GitHub’s approach to building an agentic LL model or structure for testing for accessibility. And it goes it walks through kind of their initial planning stages, talks about why they decided to go with an orchestrator model using sub-agents, which is important. and they call out why they specifically did this. Talks about some of the things that they found in their testing. it’s just a really nice overview. I wouldn’t say a deep dive because there were definitely questions that I had kind of thrown out. I’m like, ooh, did they do this? How did they approach that?

Mark Miller: Mm-hmm.

Justin Stockton: but it gives a really nice overview of how they built what they ended up building, why they built it the way they did, some of the trade-offs, and some of the things specifically that they are looking to evaluate as they continue using this new tooling. I didn’t see anywhere where it has been fully launched yet, but they have done a number of different installations in a number of different repositories. They have a good feedback model from developers that are using it. and it looks to be just a really nicely developed tool that is going to help GitHub and people that use GitHub and their tooling to be able to actually utilize agentic AI to do some accessibility testing where it makes sense is the important part.

Mark Miller: So can you talk to just like specifically and it’s so the article that Eric wrote, that is it’s sort of a making of, right? He talks about the process behind creating these tools. Let’s take a step back and talk to me about what these tools are. So we get that it’s agentic AI that can, you know, test for accessibility. But let’s get into some of the details of it. Like how does it work? What exactly can it do? Can it not do? and how would somebody really use these tools at all? I’m assuming everything’s gotta be in GitHub, but how would somebody use these tools in GitHub or outside of GitHub or whatever?

Justin Stockton: So the way that they mention in the article that it’s being integrated into GitHub, I’m assuming through Actions, but also you can use it through their copilot CLI, which is a command line interface, and or VS Code, Visual Studio Code. They have a copilot GitHub integration there as well. So those will be some ways that you can interact with this, with this LLM.

Mark Miller: Mm-hmm.

Justin Stockton: the example that they show in the article, they talk about in the article is within pull requests. So a pull request is I do some work, I’m done with that work, I’m now going to submit that work for inclusion in our code repository. And you, Mark, are going to be the reviewer of that code. That’s typically a very manual process. And you can go through and look for all sorts of things. But one of the cool things that they do they’ve done is they have a by tying this into a GitHub action, they can actually have their accessibility LLM actually be one of those reviewers. And so it can look at the breadth of the code and the action has an understanding of all the changes that were made as part of that particular pull request. But it also has access to the code. So it can look to see, hey, you know, you’ve made this particular change here. I don’t think this is the right thing. So you need to go and maybe update this particular CSS class. Or you should, you’re gonna run into a particular issue with meaningful sequence because of the way that you’ve changed this code in this way. So it can provide you feedback as part of the pull request that you could then, the developer would then go back, fix, submit changes to their pull request, you know, before that code ever gets merged in.

Mark Miller: So I really one of the what I really like is the way that you phrase that because when I was reading through this, one of my, you know, one of the things I was I thought is like, okay, well, doing these pull requests, you’re gonna have this AI agent going in and reviewing your code. The danger being that that’s what people rely on. People are like, I’m all set with the accessibility, right? This AI agent’s reviewing it, and I’m gonna do everything it says, and we’re gonna be in good shape. But you phrased it nicely because you said one of the reviewers, right? and I think it’s important. and anybody in accessibility knows exactly where I’m going with this because this is the accessibility conversation, that nothing automated is gonna catch all of the issues. And in the article it’s very clear about that. It’s like, hey, this is automation. It doesn’t catch everything. You’ve got to test your stuff. Now whether you use AI as a step during those pull requests. So it’s a shift left. This is sort of looking at the code long before it hits production, you know, in the development process. And then you have a manual review down the road on something that’s close to maybe not production, maybe it’s in staging or whatever, but you have that manual process. It doesn’t necessarily need to be here. It could this could be, could be the only reviewer. You could add a human reviewer. But somewhere along the way, we all understand that this needs to be manually reviewed still. But here’s the thing: is that you and I both know that this is people, people, this is always this question mark around how do we integrate accessibility into the software development lifecycle? How do we in DevOps, how are we doing this? And this is what I like about this is this now sort of if you use GitHub, this is now being kind of fed to you. Right? Like it’s no longer a complicated integration. so to speak. You just have to use the tools, which I think is really cool. And particularly if that brings awareness. Like we have created this step and we’re starting to understand accessibility and we’re starting to fix as many issues as we can while where it’s inexpensive versus wait till production where it’s expensive. But just by sort of GitHub serving that up and saying, We care about accessibility, here’s a bunch of tools for you to use, hopefully that also increases awareness and gets that manual check in there somewhere and, you know, all of that.

Justin Stockton: Well and the thing that and the tasty center at the middle of this Tootsie Roll Pop that Eric has created that is this it has well maybe two. But what he but what I liked is that he talks about he talks about using an orchestrator model for the agents and talks about how they’re sandboxing and that’s great. But

Mark Miller: Three l it takes three legs to get to the set that one you talk about. One, two, three.

Justin Stockton: and that’s well, I didn’t mean to be dismissive about that, but the really important part is that they’ve tr they came to this conclusion of needing to split this up and what they needed to do because an LLM off the shelf, Claude Code, Chat GPT, whatever it is, they’re not going to do, they’re not, you can’t just say like make sure it’s accessible. You really have to hand hold it. You have to put that thing on, you can’t.

Mark Miller: Wait, what you c you can’t do that?

Justin Stockton: I mean,

Mark Miller: So all the time I’ve been just typing into A AI make this accessible, that’s not working.

Justin Stockton: But behind but you have to put it on guardrails. And what they’ve done is they’ve they looked, they have a whole repository within at GitHub where it’s the results of all their audits. And they’re all contained within a repository. It’s great content. It’s great. It uses standardized metadata for describing things, for providing remediation values. So they use that information to help.

Mark Miller: Okay. But I had to figure it out.

Justin Stockton: One, train their model, but two, put their agents on guard on Rails. and that’s really, really important. That’s really cool. that’s what you know, those of us that are proponents or excited about AI and agentic LLMs or agentic models, that’s one of the things that we’re starting to see and what and the power of that.

Mark Miller: Mm-hmm.

Justin Stockton: so there’s some really cool work that’s being done kind of throughout the accessibility industry. Carl Groves has an AGP, an ARIA a set of things based around the ARIA what’s the word I’m looking for? Pattern library. that that he wrote that could be really interesting way to help guide an agent. Charlie Triplett has his stuff around that could be really useful and

Mark Miller: Mm-hmm.

Justin Stockton: Doing an agent. So there’s some really cool things like ideas and things that are floating around in the accessibility space around how do we put these agents on rails so that they’re not thinking for themselves and they’re not acting as a human, but how do we kind of start to play around in that space between the two?

Mark Miller: Right.

Mark Miller: People want to, present company included, think about AI as this anthropomorphic thing on the internet, this human-like thing on the internet. Because if I grab my phone and I turn it on, or or or if I use it on my, I can talk to it like a human being, right? But what we know in the workplace is that it’s a tool. And what I like about all of this, Carl Carl Groves ARIA, the the what we’re talking about here inside of GitHub, is this is people going, okay, we’re not dismissing AI, but we’re not pretending it’s our a friendly little robot that’s gonna follow us around like R2D2, right? We’re admitting that it’s a tool. And what we’re doing is we’re trying to figure out the best shape of that tool to serve us. So it’s kind of like if you were like, hey, look, you can, you can bang on sticks with something and it drives it into the ground, right? And you you pick up a rock, you pick up a log, you pick up your friend’s head, you know, and you start banging, you start banging things into the ground. And then somebody comes up and goes, like, well, hey, look at this. And you go, what is that? You go, this is hammer. You know, so it’s a stick and the heavy things on the end of the stick. And this is really the best shape of this tool. It’s better than grabbing a rock.

Justin Stockton: Hopefully not.

Mark Miller: To bang the stick into the ground, right? And that’s kind of what we have with, I think that’s where AI is, is we all are just picking it up and banging it up against things and thinking like, this should work because it’s kind of human-like. But these folks are saying, no, we’re going to put the time into understanding how this tool should be formed so that it’s as useful as possible. And to your point, now we have a hammer. And that hammer bangs things into the ground. It doesn’t cut things in half. It doesn’t do anything else. It bangs things into the ground. If we want something that cuts things in half, that tool would be shaped differently. We need a different tool. And that tool may be an AI tool, but it’s a different tool. So that’s where I’m real that’s where I really like this work. And then I think this exposure that Eric did here of the process and the thought process and all the behind the scenes work that went into.

Justin Stockton: We need a different tool. We would use

Mark Miller: Creating the right tool, building the tool in the right shape. There’s a lot of value in that because I think that that’s where a lot of organizations are. And AI aside, just trying to figure out how to do anything. And, you know, I do a lot of work with the maturity models because I’m the on the working group creating the accessibility maturity model that’s under the W three C and that’s what a lot of organizations are looking for that now. How do we understand our gaps? How do we understand what to do? How do we put this together? And this is kind of like a case study on how these folks did that. And not that it’s gonna completely apply to somebody else, but it’s going to expose how that thought process went. And I think that that’s just really, really valuable.

Justin Stockton: We need more people to show their work. and GitHub has always been it

Mark Miller: Yeah. As much as I resisted that, you know, like I d especially with long division, I don’t want to show my work. ‘Cause I did it on a calculator. That’s why I didn’t want to show my work.

Justin Stockton: But I mean it it’s GitHub and they have a history of open source and being open. Yeah, yeah. And so yeah. And so it’s awesome that they’re showing their work on this because there’s so many people that are talking about this, but we’re not showing a lot of work.

Mark Miller: This is what GitHub is. It’s just showing your work.

Mark Miller: In inversions, by the way.

Mark Miller: Mm-hmm. This is showing the work. That’s a good way to phrase it. That’s a good way to phrase it. And that is what GitHub’s all about. It’s literally GitHub is showing your work version after version after version after version. All right. Well, check out Eric’s article. It’s a really, really interesting. even if we got to we just and cut into the technical weeds here with it, I think you’ll still find interest in Eric’s article because it’s it talks a lot about that process.

Justin Stockton: Mm-hmm.

Mark Miller: and then certainly if they’re valuable to you, those GitHub tools are very interesting. Now let’s look at this structural workbook for AI accessible design systems that Anna E. Cook put together. and I have to admit that I did not you went through and read this entire book that she put together, this design systems books. I did not do that, and it’s pretty expensive.

Justin Stockton: Workbook.

Mark Miller: So the this is kind of a how-to guide, right, for using AI in your design systems. Is that the simplest way to to train this book?

Justin Stockton: Not necessarily a how to guide, but it’s well and she says this in the in the book in the workbook that it’s it’s it’s around governance. How are you going to govern AI as it and the implications it will have on your design system and the accessibility of your design system? So she’s really kind of looking at it through that lens, much in the same way, like when you and I talk.

Mark Miller: Mm-hmm.

Justin Stockton: with customers about the maturity model. You know, very kind of very similar. This is built out across six different questions, things like design bias, output ownership, user signals, ownership, who’s gonna own the output of the from the LLM. and it gets into you know a series of questions around each of those to

Mark Miller: Mm-hmm.

Justin Stockton: Essentially give yourself a s essentially score your organization and your approach and then gives you some you know some paths forward. But just the questions that you’re alone that you’re being asked and every time that you might look at it and be like, we’re not doing that. Oop, we’re not doing that, like it helps you to I always come back to like buying a home. Like when you the first time you ru you buy a home, you unless you knew a realtor, like you don’t know what that process is going to be like. Even though a realtor is going to walk you through, you don’t know what questions to ask or what questions not to ask, like when you’re going through, you know, a home tour. and so that’s why I love playbooks like this, because they put all the questions out where you can read through them and you can actually evaluate yourself, your organization, whatever. Go through, do that kind of self evaluation, but now you know the other questions that you should have been asking yourself all along and you can begin to formulate responses to those. How are we how are how am I, how is my organization going to address these things?

Mark Miller: yeah.

Mark Miller: It’s s so it’s liter I mean it’s it’s a way to uncover gaps in the process that if you don’t uncover them up front you’re in trouble after you you could be in you could be in some trouble after you’ve implemented right. Or you could be less efficient or whatever the whatever the case is, right? what what it I mean like so you you went through this this whole

Justin Stockton: You could yeah, you could you could you could be in

Justin Stockton: Yeah, yeah, yeah.

Mark Miller: guide here, like what is some of your favorite or I would say like what are the outstanding things? I think like when we apply the maturity model, one of the things that I notice from a customer is when you ask a question and they’re like, right, that’s why we hired you to do this because we wouldn’t have thought of that, right? Like that’s because and then they and they always are like, well, we were thinking about that and this is, you know, so they kind of like do get some of those questions right, right? Because ’cause everybody’s sort of smart and thinks about those things, but there’s always those guides. So what would you say were some of the things you read through this year like, I don’t know that I would have thought of that before I picked up this guide?

Justin Stockton: I thought you were gonna take that question in a slightly different way.

Mark Miller: in a slightly different way. Be like a politician. Just answer the question the way you want to answer it and don’t direct it into the question. And then you run for office. I wish you would. I’d vote for you.

Justin Stockton: Ha ha.

Justin Stockton: that’s a good point. And then I’ll run for office and well, all right, we’re not gonna go down that rabbit hole on a Friday.

Justin Stockton: I so the kind of the two places the the two sections that I really liked were the first section around design biases, and just that just the very pointed questions that help you to hopefully putting that in air quotes hopefully recognize that you may have made some assumptions in the creation of your design system that may or intentionally or unintentionally affect persons with disabilities. and so that’s an that’s a

Mark Miller: So un unintended consequences of dis of decisions.

Justin Stockton: Yes. Like did you accident did you unintentionally introduce some design biases that are going to exclude people? and here’s some ways to think about that. But mostly I also like that she included neuro neurodivergency in there. Neurodivergency? Yeah, that’s a word. and so so that’s included in there.

Mark Miller: That’s really cool.

Mark Miller: Neurodivergence, yeah.

Justin Stockton: So I really and I like that that is up front. It’s kind of like, you know, the Bill of Rights, you know, you know, the having it’s Friday, my brain just went away. It just went

Mark Miller: It’s sort of a mantra. It’s sort of a mantra it’s it’s it it informs sort of all the other

Justin Stockton: It sets the stage for the rest of the document. Like, and then but I also the one that I that I really liked also was output ownership. because I believe on following through and that you can’t put these things out there and then just let them be. The like when I do when I do work, like every I don’t know. Probably every couple of weeks I’m reviewing my agents files and my architecture files when I’m building something with with Claude. and so I’m constantly going back and reviewing those things, making sure, is there a better way to do this? Should I be doing something different? And so section three about output ownership, naming someone, not the AI team or the design team, but putting a name on paper of

Mark Miller: Who is going to be responsible for going back through and reviewing these things? making sure that you’re capturing the the information, that you’re logging what’s being asked of the AI and what it’s doing, making sure that and that someone that there’s an audit trail of that. Like that’s all great stuff.

Mark Miller: It goes it goes it goes back to shaping the proper tool and not just kind of let it run. And I I love the term ownership. And I’m not so sure everybody thinks about that term because ownership means I own something, right? This is mine, right? But when you use it in this type of a context, it it does it means you own it and therefore are responsible for it. And are responsi it in your

Justin Stockton: Yeah.

Justin Stockton: Mm-hmm.

Mark Miller: You’re responsible for it to an extreme, meaning you can’t you can’t blame it on somebody else, right? And in this case, you can’t blame it on the AI. There’s an admission, I think, that the going back to what we were talking about before, like AI is a tool, it is not perfect. You have to make sure that what it output is correct.

Justin Stockton: Mm-hmm.

Mark Miller: And then going back to that mantra, right? Going back to that sort of baseline setting the stage, you’re the one who has to make sure that that AI didn’t come up with consequences unintended, right? For people with disabilities. So ownership is a really in my opinion, is a really good output of this, just that concept of a human needs to have ownership over the final output in that extreme.

Justin Stockton: Mm-hmm.

Mark Miller: so cool. This is jump to the website. There’s a little bit of information on the website, but then this is a Word document that you can download, which I really like and I really like that format because it means that you can not just read the guide, but you can also kind of apply it and take it apart and put it into contexts that make sense for you and for your teams and and all of that. So this is really

Mark Miller: I I love the edge work like this. And I think if we had to maybe this this that’ll inform the title of this podcast, but there’s all these overarching principles of accessibility that we all talk about. And then sometimes the real aha moments are in this edge work where somebody says, I’m going to take a look at this specific thing and think about how this should operate. I know these things are both AI, but you know, that aside, that’s accessibility. That’s that’s avoiding those unintended consequences. That’s all that kind of stuff. So

Justin Stockton: There’s a and there’s a lot of overlap.

Justin Stockton: Yeah. No, there was a lot of overlap between these two, like the g the stuff with going on in GitHub and then this document. Yes, one’s coding and one’s design, but that the way that they over the way that I saw them kind of overlapping is around the policy the the the not the policy, the the process. And this is how we’re going to build something. We’re gonna we’re gonna think about accessibility when we’re when we’re coding and we’re building out agents.

Mark Miller: Mm-hmm.

Justin Stockton: to to kind of watch our back and to help us make sure that we don’t stumble. and then Anna’s document is this is how as an organization we’re gonna watch out for the design side and make sure that we don’t forget as we’re applying AI to our designs that we don’t forget about accessibility and that we’re that we’re you know we’re we’re staying up to date on it. So

Mark Miller: Mm-hmm.

Mark Miller: Yeah. How we’re gonna shape the tool, how we’re gonna narrow the tool’s purpose, and how we’re gonna make sure we understand we we put safety settings on and we also are gonna train the people using the tool on how to use the tool safely. It’s it’s a really like it’s like you’re training people to be carpenters, right? This is what the hammer does, this is what you’re

Justin Stockton: Not just banging nails.

Mark Miller: Impact driver does, this is what the saws do, this is how you operate with them safely, this is the regulations, this is what our company says you need to do to be safe, blah blah blah. I love it. Okay. So go ahead, you have something else.

Justin Stockton: I was just gonna say and also too, for anyone that actually does pick up this playbook and use it, Anna is looking for feedback. she says so on our website and in the document. So this is one of those playbooks that I I’m I’m kind of looking for an excuse to be able to like work with a customer and and utilize this. I just haven’t gotten a chance to use it. Like I see the promise of it and

Mark Miller: Mm-hmm.

Mark Miller: Yeah. Well, and that’s great because that means she’s gonna continue to iterate and improve based on the wider it’s it’s like releasing a tool in into beta or whatever. I mean a app into beta or whatever. You’re gonna have a wide range of people helping improve it. Fantastic. Okay. So the last thing I wanna talk about, and this is this is where we’re moving off and into a whole other area here, but this is the accessible medical diagnosis equipment rule to take effect this summer.

Justin Stockton: Yeah.

Mark Miller: And this is this is an HHS rule. So this is section five four HHS. The the article mentions title two as well. but what’s interesting about this is the application of it. So we think about digital accessibility all the time, right? We think and then we then that moves into kiosks, or I like to refer to them as enclosed systems. except for the other day, we were in the Pittsburgh Airport.

Mark Miller: And we had this great I know you’re like, what tangent are you just jumping off on? I know, listen to this. This is cool. I guarantee I promise I’ll bring it back around to this, to the accessible medical diagnosis equipment rule to take effect the summer article. but I wanted to so we we had breakfast in the airport. Great great waiter that was helping us, that was working with us and everything. And I’m like, hey, can I just pay now?

Justin Stockton: I was like, I was not in the Pittsburgh Airport.

Mark Miller: Do you know what said to me? He said, Yeah, hang on a second, let me go get the kiosk. And he was referring to the payment system. So I love the fact that in the general population, the term kiosk no longer means some big installed thing sitting in the middle of somewhere that you walk up to. We’re starting to understand that kiosk are these handheld devices in an enclosed systems. So enclosed systems, hardware, software, enclosed, oop, I just punched my mic, enclosed together.

Mark Miller: Right. I don’t know. Looks like a little punching bag. I’m gonna hit every once in while. so the this medical diagnostic equipment rule, this starts to look at not just the medical system. So the way that we think about it a lot is like, you know, your patient portals or that registration device when a patient walks in to the lobby for their appointment and here I am and here’s the additional information you need and

Mark Miller: Whatever, right? Like these are the things that we think of in the digital accessibility world. certainly anything on their public facing website or any of that patient-facing, you know, tool portal application, the video medical video tools, you know, where you might have your appointment over all that kind of stuff, right? But this is starting to talk about the actual medical equipment.

Mark Miller: the stuff that once you get past all of those things we think about that the doctors are using.

Mark Miller: And the accessibility of that equipment. And I think one of the interesting things that comes out of this is like how do you make an X ray accessible?

Justin Stockton: Well yeah.

Mark Miller: Right. But but these medical devices nowadays are large, large, I shouldn’t say large, but they’re kiosks. By definition of an enclosed system, they’re kiosks. So it starts to raise the question, like for me, there was this immediate blurring of these lines between what we typically think of that needs to be accessible for the patient and the questions this cause may cause me to ask more questions than than than it gave me answers. But the questions are where are the lines and when you start to think about accessibility, what needs to be accessible? So X ray, right? Does the X ray machine need to be accessible for the technician?

Mark Miller: Does the output need to be accessible to the patient? I wanted I just got an X ray of my foot ’cause I have a repeated use injury in my foot. So I’m hopefully healing.

Justin Stockton: Yeah, I read this more that it’s more for the patients and less for I didn’t get a chance to read the full rule. Yeah.

Mark Miller: No, I understand that that that’s what the article’s for, but I’m sorry that’s what I’m saying is that but but if you had said if you had said x ray machine to me before I read this article, I would have been like, Well, yeah, we should make those accessible so that the techs can use them accessibly. You know, so that we can have so that’s if techs have disabilities they can do an X ray, right? But this article is now going, No, no, no, the X ray machine needs be accessible for the patient. And I think it was t I think it’s

Mark Miller: Talking a lot about the physical accessibility of the X-ray machine, right? But it’s an enclosed system. There’s hardware software components to all of this stuff. So it starts to open up that question of how do you look at these things from that standpoint of accessibility?

Justin Stockton: Yeah. What’s interesting is some of the some of the examples they give in the article that we read. one it it’s I’m gonna go off topic a little bit here, but one of the exam things is you know, exam tables must be height adjustable.

Mark Miller: Mm-hmm.

Mark Miller: D I never d just why do you go off topic all the time? I never go off topic. Why do you go off topic? Are you gonna talk about a waiter in a Pittsburgh airport? Okay.

Justin Stockton: No. there’s like exam tables must be height adjustable and there’s, you know, things there. scales must be wheelchair accessible. And the first thing that popped into my head was about a month ago I took our dog to the vet. And you it used to be we would walk in, the first thing that they did was they wanted to wear. they would have her hop up on this little scale, and they would wear and it’s always

Mark Miller: Mm-hmm.

Mark Miller: Yeah.

Justin Stockton: Tricky to get her to s to stay, and then they would take us back to a room. This time they took us just straight back to a room. And she walked in and the exam table, they lowered the exam table down. She stepped up, we got her up, put her up on it. It weighed her, gave her that, and then they raised it back up. And I was just like, whoa, that’s like you’ve got now this one kind of piece. Now it’s more moving parts, more opportunity for.

Justin Stockton: breakage, that sort of thing, and maintenance. But it’s neat that this one unit is now serving as both an exam table, weight, or weight, a scale, but it’s also adjustable so that for bigger dogs, littler dogs, like you can adjust it for whatever the vet needs to look at. And I started, so when I was looking at these technical requirements, I was like, ooh, I wonder if, you know,

Justin Stockton: newer hospitals because you can’t just run out and buy a new one of those things for to replacement. But is that is something like that available? Like where they’re building the scales into an exam table or there’s something that allows for you know for that reuse between these types of tools.

Mark Miller: Yeah.

Mark Miller: That’s actually a really interesting thought. Like, if you think about a veterinary practice versus a medical practice, for

Mark Miller: There’s a lot of accommodation and variation in in a veterinary practice because and they say it’s hard to be a vet, right? Because they they’re dealing with different types of organisms where a doctor is dealing with humans, right? but because of that, their equipment is more adaptable. And it’s so it’s an interesting thought exercise of like what could we learn about adapting equipment for humans, particularly with this this HHS rule and accessibility.

Mark Miller: What could we learn about adapting equipment for humans from veterinary practices and their long practice of having to adapt their tools to varying organisms, right? Like, how do you if you need to take the temperature of a cat, a dog, a snake, a horse, a pig, right? A sheep? What I mean, if you need to take care of a horse, you need to get in your car. If you need to take care of a dog,

Justin Stockton: Yeah.

Mark Miller: The dog’s gonna come to you. I mean, just simple things like that start like that. So now my piece of equipment needs to work well inside of my office, but I also need that or another piece of equipment that I can take to the farm. Right? Like, I mean it starts that basic, but now, you know, I can put a I can measure the temperature of a horse one way, but I probably can’t do that for a gerbil, right? Like it so Yeah.

Justin Stockton: Probably very different instruments.

Mark Miller: And so it’s it’s not to it’s not to say that the accommodations that we maybe maybe we can’t steal a whole lot from the specific solutions. But going back to the other two articles that we talked about, right? It is about the process in which these things are looked at and considered and everything like that. It would be interesting. You almost would be you almost want to take a vet and be like, you know, you innovated a lot in your practice. Here’s

Justin Stockton: Mm-hmm.

Mark Miller: Here’s can you talk to this group of people with disabilities and about their needs and then consider how we might adapt our equipment to it? So I mean I’m not I’m just thinking about that. I just think the bringing up of that was a really interesting and seeing what they’ve done is really interesting. But I think what this article made me, you know, think about is that blurring of those lines. So I wonder in the future how these medical devices that

Mark Miller: Are being used inside of a medical practice by doctors and nurses and all sorts of medical practitioners. I think more and more there’s going to be patient-facing aspects of those that aren’t just physical but are also digital. I don’t know why, but I think of it like if I step up on a scale at my doctor’s office, I want to know what the scale said. I want to be able to look well, but I can look at the scale.

Justin Stockton: I hate it when they have it in kilograms. Yeah.

Mark Miller: Right. I have a vision. So I can look at that scale and I can go, look, I’m a hundred and sixty-five pounds. And I go, that’s good to know because now I know my scale at home is accurate and blah, whatever. I don’t know why you I just want to know. And I have the right to know, right? But if I’m if a blind person steps up on a scale, I’m sure that the technician would be perfectly happy to tell them. But what’s the equivalent?

Mark Miller: What’s the accommodation so that that blind person can let’s assume that they don’t have that human there to help them, right? Because it’s equivalent experience. What’s the equivalent? And just on something as simple as a scale. So it just makes me think about

Justin Stockton: Yeah. You would need a check because you wouldn’t want to broad you wouldn’t want it to speak to you. You can broadcast your weight to the you know, anyone’s nearby. Yeah.

Mark Miller: Yeah, right. There’s other people around, it’s a HIPAA violation if the thing starts talking. Maybe it’s even a HIPAA violation if if I can look at it and know what it says. But now when a technician blurts it out, maybe I don’t want that as a blind person. I want the ability to understand it. So maybe I want to plug my headphones into the scale. I don’t know. You know, I’m speculating. But and then also I think that when we we look at like Waymo’s and stuff in cars that are becoming autonomous, like this age of autonomy in terms of equipment.

Mark Miller: So equipment that used to be have the need to be driven by a person now doesn’t, or and we’re starting to move more and more towards that. So when I walk into the clinic ten years from now, am I gonna be sort of escorted around by a human being who’s gonna strap a cuff on me and stick a thermometer in my mouth or p you know, hip do whatever they all that stuff that happens at all? A human being is like doing that. Or am I gonna walk into a room and sit at that

Mark Miller: some version of that table that you talked about that your dog stepped up on, and that table’s gonna take my weight and a cuff’s gonna come around and take my blood pressure and something in the room’s gonna know my temperature and and maybe it’s gonna ask me questions and I a a answer the questions and so there’s gonna be is there is there going to be an interaction outside of that human interaction that we have now in the future. And if so

Mark Miller: These rules, as they should, are gonna require them to be accessible. So, anyways, it just that’s that’s that’s what I found is that’s why I wanted to bring up this article is because I think that we need to be start to really think outside the box. I think our the wor the digital world and the physical world and the world of human beings is all collapsing in together to become something different.

Justin Stockton: Hopefully not a black hole.

Mark Miller: That the that would be terrible. Is that what you’re saying?

Justin Stockton: Ha ha.

Justin Stockton: Mm no. I’ve just been reading too much.

Mark Miller: I’m gonna cancel my AI accounts right now.

Justin Stockton: Ha ha ha.

Mark Miller: Yeah, I don’t I d I don’t I hope it doesn’t yeah. Hope the wave packets don’t collapse into a black hole. You’ve been you’ve been reading Stephen Hawking lately. Is that what’s going on? Cormac McCarthy. maybe it’s Brian Green. I don’t know. I’m not putting those in the show notes. You guys can look that stuff up. All right.

Justin Stockton: Yeah.

Justin Stockton: No, that was Cormac McCarthy.

Justin Stockton: Are we to the point yet where we get to take wait? Are we to the point yet where we’re going to take bets on whether or not this ruling will be pushed? Or is that too apropos?

Mark Miller: Hundred percent it will.

Mark Miller: This is that folded up envelope. I d if I it’s got to and I was confused a little bit about the ruling. It looks like it’s it’s all five four HHS, but it looks like it’s a separate and they for some reason they mentioned title two in here, but it doesn’t look like it d it’s a d they’re different different it’s different rulemaking, I think. but

Justin Stockton: Mm-hmm.

Justin Stockton: I’m gonna go make some polymarket.

Justin Stockton: Yeah.

Justin Stockton: Probably a different rule making for physical devices, just like we have like like the ADA and the ABA and you know.

Mark Miller: Well in right. In so so July eighth, twenty twenty six, and August ninth, twenty twenty six, practices must comply by July eighth, twenty twenty six. The compliance deadline for government entities is August ninth, twenty twenty six. So given that we’re looking at probably sometime at the end of July, beginning of August, my prediction would be that the government one would be pushed and then

Mark Miller: Beginning end of June, beginning of July, the compliance deadline for practices. That would be I’m gonna fold up my prediction, put it in an envelope, mm, seal it up, and that’s what I would that’s what I would say. And and that’s based on HHS following the HHS rule following what the Title Two rule did earlier.

Justin Stockton: Yeah. Following what the DO did.

Mark Miller: And if you all are not like, what are these guys talking about? You can jump back a couple of episodes. We’ve got some blog articles. You can Google it. There’s a lot of information out on these rules being extended. and we won’t talk about how we feel about it. You can go to previous episodes of this podcast where we go in depth and talk about those things. How’s that? Forty four minutes and eight seconds. We done? We done. We done did it.

Justin Stockton: We’re done. Call it.

Mark Miller: All right. Well thanks, man.

Justin Stockton: Yeah. Always a pleasure.

Mark Miller: Thanks for the thanks for driving the technical side of this one. I appreciate that. Cause I was like, I’m not gonna get on and start talking about all that stuff ’cause I don’t have you have the right brain for it.

Justin Stockton: Yeah, isn’t it?

Justin Stockton: Good articles this week. I wish I wish it wasn’t Friday afternoon because I was how did I mess up First Amendment?

Mark Miller: I never thought I’d hear you utter the phrase I wished it wasn’t Friday afternoon.

Justin Stockton: Well, good point.

Mark Miller: On that note, I hope you have a good weekend. Now that we have broken down accessibility for you, we hope you go forth and keep it accessible.