Welcome to the thirteen episode of Access On, the National Federation of the Blind's Technology podcast.
Episode
Listen to the thirteen episode of the Access On podcast (Browser).
Or listen on your preferred podcast platform.
Timestamps
The show is segmented by chapter, making it easy to move between segments of the podcast if you have an app or player that supports chapters. Below is what's on the show this week, and when you can hear it.
- Introduction 0:00
- Your final reminder to register for our AI prompt engineering boutique 0:59
- Follow-up from some items on previous episodes 2:56
- Mike Calvo and Matt Campbell from Pneuma Solutions discuss Scribe for Documents 5:40
- When we can't choose the tools we prefer due to inaccessibility 43:57
- Access ChatGPT via phone 56:28
- Closing and contact info 59:24
Transcript
Speaker 1: Live the life you want.
MUSIC: Access On.
Jonathan Mosen: Welcome to another edition of Access On the technology podcast of the National Federation of the Blind. This week, when should we settle for a document that's just okay and when should we demand perfection? We look at Scribe for Documents from Pneuma Solutions, what it can do and when it might be the right tool. We've come a long way, but too often we're still required to choose the tool that's most accessible rather than what's optimal. And tell your friends who don't have a computer yet, they too can access ChatGPT over a standard phone line.
It's Jonathan Mosen welcoming you to episode 13 of Access On, recorded here at the Jernigan Institute where it is absolutely blustery outside, blowing quite the gale and it's pretty cold on top of all of that as well. This is the final time really I have to give you a reminder about our upcoming boutique, which I'm sure there'll be plenty of interest in. This is on How to Engineer an AI Prompt and it's happening on February the 25th from 2:00 until 3:30 PM Eastern Time.
AI tools are becoming ubiquitous in our lives as technology rapidly advances. To get the most out of these tools, it's important to ask the questions the right way. This practice of properly communicating what you want to an AI model, called prompt engineering, is crucial for making sure you get the information you're looking for as efficiently as possible. So if you join us on February the 25th, we will cover why prompt engineering is important, how to craft a good prompt, the effects of different phrasing on AI responses, real examples of how this can be useful in daily life.
Time is really running out now, so be sure to register for the How to Engineer an AI Prompt boutique today. It happens via Zoom, so that's something that many people are familiar with. It means that you can join online or via the phone even if you want to do that. And Paul Hopewell, who is in the UK, wrote in to Access On at nfb.org to say he was delighted, delighted and surprised he was that he was able to register for the How to Engineer an AI Prompt from the UK.
He was expecting that you might have to log in some way and identify as an NFB member. No, it's open to anybody anywhere in the world. You are welcome to register and attend this webinar no matter where in the world you live. To find out more and to register, head to nfb.org/cena and choose the link for the boutiques and trainings. Nfb.org/C-E-N-A is where you register. And also be sure that you don't miss out on hearing about any other webinars and boutiques that are coming. We've got a great calendar of events happening in 2025. Before we get onto our feature interview for this week, I wanted to follow up on a couple of past items on Access On.
You may remember that I reviewed for you the PLAUD NotePin. This is a tiny gadget that you can attach to yourself in any number of ways and it records any meetings that you attend and then takes notes for you. It can give you a full transcript or it can give you summarizations.
So it's kind of like many of the features that you see in these conferencing tools like Zoom and Teams, except it's for in-person meetings and I discussed some of the accessibility issues with the software that powers this device, both on the web and certainly in iOS. Another listener took up the case too and has been having some good dialogue with PLAUD about this. They have now actually responded and they've said they have listened to the episode of Access On where we demonstrated some of the difficulties that this software has from a screen reader user's perspective.
They now understand what a useful device this is and how lack of accessibility is letting it down, and they've indicated that they've added accessibility to their list of requested features and hope to get to it. Now at the National Federation of the Blind, we've pushed back a little bit on this and we've said, "Actually if an app is designed from the get-go correctly, then it would have been accessible." So the fact that the app isn't accessible is actually a bug, it's not a feature. We don't consider adding accessibility to be a feature request.
We consider it to be addressing a deficit in the existing code. In our view, it's a critical deficit in the existing code and we hope that it will be prioritized accordingly as a critical bug. So the good news is the dialogue is taking place and that we have been heard and it would be great to see the software for this very useful device improved in future. We will certainly keep you posted. Similarly, some episodes ago, Asia wrote into us.
You may remember that Asia said she had a new job and she was having some difficulty performing some of the core functions of that job because of some accessibility problems she was experiencing with Microsoft Bookings. We reached out to Microsoft and we said, "Look, we've got an NFB member here who's got a job. We know how difficult it can be for blind people to get those and now the problem is the technology's not cooperating as it ought to." Microsoft have been very responsive on this point and assigned someone to work directly with the listener, Asia, on this. Microsoft has kept us appraised of developments.
They say they have identified a defect that was affecting Asia's use of the product. Effects is now in testing and it will be rolled out sooner rather than later. So that is a great outcome and an example of the difference that the National Federation of the Blind and this podcast can make. Document remediation is something that is of particular concern to many people. You find a document that you really want to access but it isn't accessible.
It could be completely unusable or it could be that there are just not headings in the right place so it's difficult to navigate. There are a whole range of document remediation problems, so it's a hot topic. What is also a hot topic though is whether automated solutions can work. Just in last week's edition, we spoke with President Mark Riccobono of the National Federation of the Blind about accessiBe and ongoing concerns about remediation of websites. So can documents be remediated this way effectively? Well, Pneuma Solutions has been operating Scribe for Documents for some time now and they've recently announced some enhancements to that.
And I thought we would talk with Mike Calvo and Matt Campbell about this and also about the concept of whether automated remediation really can work and save companies significant amounts of money while also delivering quality products to those who need them. So Mike and Matt, it's great to have you on Access On, welcome.
Mike Calvo: Happy 2025. Welcome to the US, man.
Jonathan Mosen: Thank you, it's good to be here. All right, let's talk about this. What is document remediation for those who are unfamiliar with that term? What do you mean when you talk about document remediation? Matt Campbell we'll have you take this one.
Matt Campbell: Document remediation is the process of taking a document that is to some degree inaccessible and making it more accessible and it can be automated, human-remediated or a mix of both.
Mike Calvo: And what you're doing is you're actually taking parts of that document and segmenting them. So as we use customized commands in our screen readers or whatnot to jump headings and go to lists and go to buttons and read graphics and all that stuff, what document remediation does is it marks all those things so that the person using the AT can navigate the document. It also takes a document, for example, a PDF document that is in its 100% graphics form, which a screen reader can't read a graphic, and using OCR technologies and other things, converts it to text and then formats it in that specialized form that we just spoke about. In the case of Scribe for Documents, of course, then we can take that remediated content and then drop it into a number of different file formats. Be that Tagged PDF, Microsoft Word or accessible HTML or DAISY or Braille even, large print, MP3, a bunch of different formats.
Jonathan Mosen: One of the problems though is that people have said if you rely on a machine, some sort of algorithm to do this remediation it sort of works, but it doesn't give you the same degree of granularity that human remediation will do. For example, you might be able to see from a poorly formatted document that if the font size is raised a bit and the text is bold, you should make that a heading level. But what automated document remediation has failed to do in the past at least is give you structure. So should that be a heading level three for example, because it's part of a subsection, not a new section? And that's why people have frowned in the past on automated forms of remediation. Has Scribe for Documents got that sorted?
Mike Calvo: I think we have, but I want to step back a minute, Jonathan, because and I don't mean this in answer to what you're saying necessarily, but as a blind person in general, I'm sick and tired of hearing everyone in our community want everything 100% perfect. Life isn't perfect you all and I'd much rather have information access and if it's screwed up and gave me a heading level two instead of a heading level three, but I've got access to the content underneath, which is something that I get with augmented or automated document remediation and something I don't get with human remediation, the elephant in the room is the cost. You're talking an average of anywhere between 10 to 15 dollars per page and half an hour of human time to remediate one page of something. And I would submit to you that all documents, while their importance in your day-to-day life is not equal, 100% access to them in some form is equal or should be equal.
And when you go to a company and you're a business guy, Jonathan, when you go to a company and you say, "Hey, you know what? You've got 5,000 pages here. You've got 10,000 pages that you need to remediate." And they say, "Well, just go ahead and remediate it all." Well, wait a minute, you're going to want to prioritize which one of these documents you've got to do because it's going to be about $17 a page. So at 10,000 pages, that gets to be a bit expensive. Not to mention it takes forever and a day. With the approach we take at Pneuma is we say, "You know what? Let's create a platform that gives a pretty darn good most of the time." And we've had this debate in our community before, what's the difference between usable and accessible? And I would submit to you that 100% accessibility costs a lot of money, but you know what? 90% to 95% usability does not and that's an important differentiation because I've never seen a company taken to court for making a really good try.
Not to mention the fact that if we provide a pretty usable document and the consumer wants it in a human remediated form, in other words they want it more perfect, we provide for our customers a way for the consumer to click a button and within 72 hours they have a human-remediated version of that document. It saves a lot of money for the customer, for the content owner, and it also allows them to be able to just throw everything at the remediation platform. So that PTA meeting that happened in 1985 on September 23rd when you were in elementary school, you can go look at that but you can't do it now.
Jonathan Mosen: The consumer is not always the customer so if a blind person makes a conscious decision, "This document isn't important enough to me to care too much about structure. As long as I can read the document, that's really all that matters to me and I'm willing to pay a few pennies for that and accept the potential consequences." But isn't there a risk that because often the purchaser is not the consumer, that the purchaser takes shortcuts in documents where structure is really important. For example, if it's a complex legal document or a complex document where navigating from one chapter to another or one major section to another and drilling down into the subsections is relevant. But somebody says, "Oh, I'm going to take a shortcut here and not get this done properly because it's cheaper." Especially in the current climate, isn't there a risk in that?
Mike Calvo: I think that that's like saying AI is going to take over the world. Let's stick to science and not science fiction. And the reality is that of course they're going to take the lower cost route to begin with. But again, if you need it in a more accurate form, then ask for it. The reality is that if a library has 1,000s of documents available to it, if they take the money and they spend their money on remediating every document and that document in the entire history of that library has been accessed by 15 people, that's not good fiduciary responsibility.
However, if you give the opportunity for the consumer to look at it with a promise, with a pathway, with an opportunity for human remediation should it be asked for, it's going to happen and we can give you that document within 72 hours, which beats the tar out of just about any accessibility office in any campus in this country.
Jonathan Mosen: So can you explain how the workflow goes if somebody wants to use Scribe for Documents for remediation, what does the process involve?
Mike Calvo: Yeah, well, by the way, and maybe I put the cart before the horse here, there's a couple of ways. We can use augmented document remediation or Scribe for Documents right away. Go to the website, you upload a document, it gives it back to you in seconds. I mean it takes about five minutes to seven minutes to do a 500-page document, which is pretty darn impressive. Where we really see Scribe for Documents being huge is for legacy documents. It's when an organization has years and years and years of accumulated documents that they need to bring in.
They need to get these done. And there's this little thing called Title II of the Americans with Disabilities Act that the Department of Justice came along a couple of years ago in its great wisdom said, "Hey, we're going to give all local, state and federal governments until April of 2026, if you manage or if you govern more than 50,000 people, and up to April of 2027 for less than 50,000 people." Why that is? I don't know. It would mean I would do it the other way around, but okay.
And if you're not Title II compliant, which means making all your websites, all your documents and all your mobile apps accessible, then there's going to be basically hell to pay. Because then it opens up the market for lawsuits and for all kinds of things, and we could get political about it and say that the current political mood is not necessarily focused on us, but does an organization really want to take the risk? Do you really? Because it doesn't matter what the government says. The court of public opinion is going to judge you just as harshly, so providing access to this information is really important. On the consumer side, you just go to scribeit.io. I think Matt, do we have scribefordocuments.com up already?
Matt Campbell: Yes, that's redirecting to scribeit.io. So Scribe It but yeah, you can get there either way.
Mike Calvo: Yeah, it's easier to remember and you can pay for a monthly subscription, an annual subscription. You get 2,500 pages for that for 250 dollars and there's other things that are going to come down the pipe. But the thing that we do really well is we can do 100s of 1,000s of documents in an afternoon. So if you give us 100,000 pages of content, for example, we can do it in about a week and have it finished to you and marked up, and we can do that for pennies per page.
Jonathan Mosen: If you are using a tool like JAWS Picture Smart and you take a document that's an image that somebody's given you and Picture Smart describes that image and reads it to you, what is the value that you perceive Scribe for Documents to be adding over something like that which is built into JAWS now?
Mike Calvo: Well, of course, the first thing is that you need JAWS. The second thing is we're not that silver bullet. We are a solution in the toolbox. Our bread and butter does not come from the consumer. Our bread and butter comes from the owner of the content. I mean it's like anything. It's like Aira, it's like any of the organizations that are out there. Our community is small and our community does not have a whole lot of money, but Matt and I, we refuse to hold enterprise level features back from our consumer because of money.
So we give our consumers at a hugely discounted price what we sell to enterprises for substantially more in bulk. But basically our customer is anyone with a large content library. If consumers want to use it, we're going to be doing things like we're going to create a Scribe for Documents desktop app, and there's other things that are coming down the pike for Scribe for Documents this year that will make it a little bit more consumer friendly.
But right now I would say in your arsenal as a consumer, it's something you can add to your arsenal and it makes it very cool for you to have. But for an organization with a large, large library of content, legacy content, especially those that are dealing with Title II that need to remediate things now, that's really where we shine.
Matt Campbell: I would like to call attention to a couple of features that we do have that would be of interest to consumers that I don't believe JAWS Picture Smart or any other existing solution has. And the biggest of those, the main one in fact is we just rolled out support for mathematical notation using a third party technology that we're using as part of our workflow now. When you provide a PDF or an image as input to Scribe for Documents, whether that PDF is a scan or a PDF with an existing that already has text in it, we can process it.
And if it has mathematical notation, then we will render that as accessible MathML that you can access with a variety of screen readers, with JAWS using its built-in MathCAT integration, with NVDA using the MathCAT add-on, and with VoiceOver and ChromeVox as well. That's an optional feature that you can turn on because it's currently in beta and we do get some false positives, so we've made it an optional feature for now. We all-
Mike Calvo: We also have this Nemeth so you can convert to Nemeth.
Matt Campbell: So if you convert to Braille with the Nemeth code for the math notation. The other thing that we have that might be of interest to people that like to listen to certain types of documents with a very natural sounding voice is we've upgraded to a newer neural AI-based text-to-speech engine. So if you take a document and convert it to MP3 or DAISY, then you get very nice voices and it is in fact an upgrade from the TTS that we had before.
Mike Calvo: Also, the alt text stuff is better and we're working on that too because we want to provide extended alt text descriptions. We don't have them yet, so we're kind of tweaking the LLMs to see where we can go, the large language models, to see where we can go with that. Because a short version in the middle of your reading is good, but sometimes people really want to dig deep into it.
Another thing you could do with our document remediation platform that you can't do with Picture Smart is automatic language detection and providing that content in a markup language that makes your TTS speak in the proper language for that content. So you can have English, Spanish, Italian, French, all on one document and we will detect that and when we create the remediated output, it will be marked up properly so that your screen reader obviously, provided you have the voices installed, will speak it properly.
Jonathan Mosen: And it is about selecting the correct tool in the toolbox for the particular task in question. Because you mentioning alt text reminds me of the classic AI dilemma, which is they can describe, for example, a blind guy holding a white cane but they can't say that it's Mike Calvo holding that cane because they don't have that information.
Mike Calvo: Not yet. They do to a certain degree if Mike Calvo was Bruce Willis or whatever and he was up there, some of this stuff will describe public figures. But yeah, I mean again, look, the reality is that a sighted person can get a messed up PDF too. Matt and I were talking about it today. Sighted people bring up a PDF on their phone, looks fine on their computer. They bring it up on their phone, it's one column and they've got to deal with that. Do they have the right to sue anybody for that? No, they don't. So the reality is that it's not perfect.
Welcome to the real world. It's not perfect, but it's better than not having it at all. I'd rather have 95%, 96%, 97% of something than 100% or nothing. And by the way, I guess in our defense, and I'm not being defensive, I'm just covering the bases. We will come back for a customer and re-scan every 18 months because we believe that in the next couple of years we're going to be at pretty damn close to 100% accuracy on this stuff and we will go back.
And that's another thing, remember we talk about human remediation and how perfect it is. Humans make mistakes too and once it's done, it's done. If we remediate a document now and for some reason it has a couple of little faux pas in it, not enough to kill it. But hey, it's the little bumps in the road. We do come back and re-scan it and that will generally fix it because the technology we can assume has gotten better between now and 18 months from now.
Scribe for Documents was released years ago. It has definitely changed. I mean when we look back, Matt and I did Freedom Scan back in 2002. We did DocuScan Plus back in 2010, which is version two, and then we did DocuScan Plus version three in 2015. We've got some experience at this. We've been doing it for a long time and I think that being blind people to touch on the accessiBe thing, I like the accessiBe model. I just don't like the BS that they used behind it, but I think that it was a very noble concept, however it worked out.
I am not an accessiBe fan. I'm glad that they got their hand slapped in the way that they did. That doesn't mean that, oh, my goodness, everything that comes out and says I can do this in that amount of time is going to be the pig in the poke that accessiBe was. Especially when you're talking about two veterans in our community that we've been here just a minute and we happen to know what we're talking about when it comes to what we as blind people need.
Matt Campbell: And not just us, but we are obviously standing on the shoulders of giants in the field of OCR in particular and document structure analysis. AccessiBe is solving a much harder problem making it potentially very dynamic websites accessible than the problem that has been iterated on going back to pioneers like the Finereader Engine and the OmniPage Engine and newer ones that we're using today of making a static document accessible by analyzing its content and structure.
Jonathan Mosen: In terms of setting expectations around when this is a good tool to use, you mentioned that this is particularly suitable for organizations with a large amount of text to remediate at once. Given the state of OCR these days, what quality output am I likely to expect if it's say an archive of old newspaper material or written documents that may have seen better days, how likely am I to be able to read that material effectively once it's been through the remediation process?
Mike Calvo: I think to put a number on that like, "Oh, Jonathan, hey, we can make your 1867 newspaper 80% accessible." I think that that's disingenuous if we did that. We're going to give it a shot. It's going to work in a lot of things. I've seen some pretty marvelous things happen, but I've seen some pretty bad things too. So it's another tool, garbage in, garbage out. We can't put what's not there, but we can certainly enhance what is there and we're going to- and then remember, this is not the end.
This is okay, if you've been doing research and if you're doing research on the Salem Witch Trials and you've got all the newspapers from back at that time and all you can tell or read, all you can do by using Scribe for Documents is read the headlines in each one. Well, now you know what to point at and say, "I need this human remediated and that human remediated and the other and the other." But to say, "Well, this is all old stuff from the 1800s or 1700s and it's going to be 17 dollars a page." And if it's really bad, it's going to be more than that because half an hour of work on a page of just a remediated document is what it takes a standard remediation. Extended remediation where they've really got to work at it takes longer.
So this way at least you can look at it and then say, "I need this, this, and this." And the organization would be more than happy to do it because they didn't have to pay 17 dollars a page or whatever for the whole thing to be remediated.
Matt Campbell: It may also be worth pointing out that as it did when we first talked about Scribe for Documents on Mosen at large a few years ago, we offer the first two pages of any document as a free preview. So if you've got a document that you want to throw at it and see how it does, you can go to scribeit.io and you can give it the PDF file and get the first two pages as an HTML preview within seconds for free.
Jonathan Mosen: We've talked a lot about PDF and that's because it is a bit of a problem child with this sort of thing. Are there other formats that you will accept as input?
Matt Campbell: Yes, we'll accept image files, so GIF, JPEG, PNG. We'll accept Microsoft Word, Excel and PowerPoint as well as EPUB eBooks.
Jonathan Mosen: So this is interesting because one of the things that's frustrating about Excel spreadsheets is if you just structure them correctly and you define your header rows correctly, it's not that hard to make an accessible spreadsheet, but it is amazing how often that doesn't happen. So is that something that you can sort out more times than not if I give you a spreadsheet that could just do with a little bit of accessibility magic added there?
Matt Campbell: To be honest, Excel spreadsheets are not an area that we've focused on much. I know that we'll turn it into an HTML table, but I can't make any promises about how effectively we can remediate the kinds of nightmare spreadsheets that you've probably seen that we haven't necessarily.
Mike Calvo: Yeah, again, this isn't a silver bullet. I mean it's not going to fix everything. It's a way of getting it at content. Another thing that we're looking at doing is adding Scribe for Documents either through an extension or through deals with the organizations. How do you do, if you go to Amazon, for example, and everybody up there has PDF files for their manuals? Amazon is interestingly enough, they're liable from a legal perspective if those things aren't accessible, but they have no way of controlling that.
Jonathan Mosen: Right. So another example that I'd be interested in seeing whether you've played with was when the Zoom Essential Recorder series came out and it created a real buzz in the blind community because finally there were these accessible recorders with menus that talked and they've made a real difference to many content creators. The problem was though that the user manuals were not accessible because Zoom had little graphics and they hadn't attached text labels to that graphic.
I think that that has largely been addressed in some of the updates to the manual. But is that the kind of example where Scribe for Documents may have assisted? If we had put those Zoom manuals through Scribe for Documents, might we have got something back that would have been helpful in terms of when the manual was telling me to push a particular button, you would know what the button was?
Mike Calvo: I don't think so. What do you think, Matt? Because the thing is that you have to define those things in... So I mean if you create a document profile, sure, if we tell the platform whenever you see this, it's called that. But in its kind of out of the box state, probably not.
Matt Campbell: Yeah, in its out of the box state, the AI-based image description service that we're using on the back end might be able to tell that there's a right arrow icon here, but that doesn't necessarily, I mean it wouldn't be able to infer-
Mike Calvo: Not like a platform [inaudible 00:31:30].
Matt Campbell: Or that's the play button or whatever. That's another issue because the Play and the Fast Forward, at least in the standard icons that I've seen, Play and Fast Forward are similar icons, but it's like the Play button is a right arrow, whereas the Fast Forward is a right arrow with a vertical line. And is the generic AI image description service going to be able to distinguish between those things? So there are limitations of what this kind of automated process can currently do, but notice the word currently. AI and particular multimodal models continue to move forward and we'll continue to explore what we can do with these newer technologies.
Jonathan Mosen: You talked about a profile where you can define certain graphics. Is that something that the person who is remediating the content can set up? So it's kind of like maybe a set of instructions you might give to a transcriber about unusual words, for example, is that how that profile system works?
Mike Calvo: It's something that we can do for library content at this time, but it's not in the platform. Hopefully that will change. Right now, the big thing that people use Scribe for is just for documents. It's available on your phone. It's available everywhere because it's a web-based application, so people just upload documents and look at them.
And I can see potentially a day where we create a little something that goes in your phone that makes Scribe part of the share sheet along with Be My Eyes and Aira and everything else that we now have. Thank God we have so many things available to us, but the profiles is something that we would do right now in a library scenario, where we know that these documents have a certain idiosyncrasy, so we would put that in the platform for this particular run and not necessarily every run.
Jonathan Mosen: What do you do with PowerPoint presentations? If I feed you a PowerPoint and sometimes they are pretty graphical and not always as accessible as they ought to be. What can I get out the other end? We have talked a little bit about input, but let's talk about output. What I can get out of this? There are Word document options, I take it HTML and potentially MP3 as well.
Matt Campbell: For PowerPoint, we run it through the same engine that we use for Scribe for Meetings to convert the slides that the presenter or meeting organizer uploads there. Again, we run it through AI-based image description and we can detect things like multiple columns of text on a slide and try to rearrange it into a sensible reading order. And then on the output end, as you said, we can't output back to PowerPoint, at least not currently. But we can output to HTML and that is a single self-contained HTML file that has any images embedded in it, and then we can output to Tagged PDF, Word, Braille, MP3, DAISY, EPUB and MOBI in case you want your ebook in that format for some reason.
Jonathan Mosen: I know you've recently upgraded Scribe for Documents. Are there any other new features in particular you want to highlight that we haven't covered yet?
Mike Calvo: Imagine a world, and the reason I'm going to say is because we're still trying to play with the model, right, as far as the business model goes. That's constantly the bane of our existence is, "Okay, how are we going to make money this month?" Because we're trying to build a technology, we get pushback and as any technology company does. And we get skeptics as many technology companies rightfully get in our community so we get it.
But imagine a world where you could go and say find a manual, a user manual and right there there's a button that says, "Read an accessible version of this document." You click that button, within seconds that document pops up in another window in an accessible way, in an accessible HTML, and then at the bottom there's a link that says, "Download a digital version of this document." And you can get it in Braille, MP3, all of the different formats, including Tagged PDF and all of that is in real time.
So for an organization like a Home Depot or a Lowe's or an Amazon or a Best Buy or a Walmart, something with huge, they don't have to pay for every document to be remediated. All they've got to do is basically pay for a subscription to be a part of this network every month, and then they pay for the pages that are remediated only. And, of course, the benefit to them is if the consumer still needs the human remediation, there is a button in the platform that allows that. Does that sound like something that you would use as a consumer?
Jonathan Mosen: It sounds like something I would use as a consumer, but right now it's a product of our imagination, right? Because as we've just discussed, if we feed a user manual to these services at the moment, they're not going to be able to tell me what's the Play button? What's the Stop button? What's the Fast Forward button? So we cannot use automated technology to produce accessible user manuals yet.
Mike Calvo: That's kind of throwing the baby out with the bathwater, don't you think? I mean that's the only thing you read a manual for is to know where the buttons are?
Jonathan Mosen: If you-
Mike Calvo: You don't read a manual to know how to install it, what it goes with all the tips and tricks on, I mean that's-
Matt Campbell: Well, but if the tips and tricks... I see Jonathan's point though, because if the tips and tricks section says, "In order to do this, press the icon. Yeah, icon goes here button and then press another icon button and then press the third icon [inaudible 00:37:33]."
Mike Calvo: Right, that's when you go for human remediation.
Jonathan Mosen: Right.
Mike Calvo: That's when you pass-
Jonathan Mosen: And my argument would be that a blind person should not have to settle for that. That actually if these documents are born accessible in the first place, and we encourage these producers to do the right thing from the get-go. Because in this scenario that you are painting here, Mike, we're not talking about documents that are decades old. We're talking about documents that are produced today and they should be accessible from the get-go. That's their responsibility to produce those documents in a born accessible manner.
Mike Calvo: Sometimes we get, I mean and I understand and I agree, if Sony or name your large company puts out something, they should know better. But what about the smaller places? What about the smaller? We get a lot of imported stuff in the United States and in other places that come from different countries. And those organizations may not know about accessibility and granted they should learn, but at the same time, again, you get back an all or nothing scenario.
Should we as a community sit around and wait for that kumbaya moment where everything is accessible and advocate the hell out of things? But meanwhile, grass is growing and cows are starving. Or should we jump on what's here and continue to advocate? It's not a one or other scenario and accessiBe, and I use accessiBe because I think it's appropriate, it did not stop people that really gave a damn to remediate. The people that wanted to throw a Band-Aid on their website were going to throw it on there anyway.
So I mean people who truly understand remediation and have a commitment to their consumers aren't going to allow that to happen. But I still stand on the fact that, yeah, the Play button may not be something that's accessible but like Matt said that's today. But besides that, that's certainly not a reason to just say, "Well, then let's not do the manual at all because we can't tell what the buttons are on the remote." That just doesn't make sense to me. It's not something I would do. I mean I want to know about the rest of it.
Jonathan Mosen: But essentially if you can't tell what the buttons do and what buttons to press for specific functions, the manual remains inaccessible.
Mike Calvo: Yes, and no. Again, you're right in that the manual doesn't remain inaccessible. That part of the manual remains inaccessible. Again, if I'm looking to read about a product, there's a lot more in that manual than just the schematic and the layout of what the thing does. I mean there's information about how to use things, what troubleshooting is like, what this and that. And yes, those buttons are extremely important, but we're focusing on one thing. What about the person that needs to install a shower or change a lock in their house and there's a manual that goes with it? I mean, there's a million different manuals out there that don't have button descriptions that deserve their opportunity to be remediated.
Jonathan Mosen: It's a great discussion and it's a perennial discussion because we're often talking about what is the responsibility of the producer of the product and what are the opportunities of remediation? And I remember when JAWS added OCR into the product so that you could OCR in an accessible PDF, and there was the same discussion then that some people felt that Freedom Scientific was letting the producers of the documents off the hook. So it'll be an ongoing discussion and it will be great to get listener feedback on this.
We do have to wrap now, but I hope we will get some listener feedback on this and we can discuss both sides of this going forward. So scribeit.io is where people can go and it's pretty easy to see what it's like and you can kick the tires and put it through its paces. And it's great that you're looking at these issues and thinking about solutions, so I look forward to finding out what happens with Scribe for Documents and other products in the future.
Mike Calvo: Jonathan, thank you for a great discussion and a stimulating one as always. You're truly one of my favorite people just in general, but as an interviewer, I love your candor. I love your approach and thank you for calling us out where it needs to be done. I respect and appreciate that.
Jonathan Mosen: That's Mike Calvo and Matt Campbell from Pneuma Solutions, both blind people who've been around for a while. You may remember them from the days of Serotek and all of the products that they did there, and I would be very interested in your thoughts on this topic. How is a purchaser to know when automated remediation is good enough? Because the problem, as I said in the interview, is that so often the purchaser is not the consumer. So you may have a purchaser who tries to take a shortcut and decides that remediation done in an automated way is an acceptable solution.
When as the end user, we may think that it's not because structure is really important and those graphics and their interpretation really is important and we suffer from those shortcuts having been taken. But on the other hand, as Mike has been saying, is it better not to have some sort of access to the document than none at all? Your thoughts on this, and particularly if you've had any practical experience of this, because certainly Scribe for Documents is not the only automated remediation solution that's out there.
Share your thoughts with us, [email protected] is the email address and you can attach an audio clip or just write the email down, Access On all joined together @nfb.org. If you're a member of an NFB chapter, be sure to say hi to that chapter. Tell us where you're from. We'd be very interested in your thoughts on this. When we come back more of your comments on Access On.
Speaker 6: Join our 2025 National Federation of the Blind National Convention from July 8 to July 13 at New Orleans, Louisiana. Our National Convention is the largest gathering of blind people in the world. It is the premier event for training, support and information for the blind community. The National Convention offers 100s of sessions, dynamic presentations and countless opportunities, as well as a chance for members to participate in decision-making in organizational policy. Now is the best time to book your hotel stay, learn more at nfb.org/convention.
Jonathan Mosen: This email is a long one, but I want to read it in its entirety because I think it raises an important subject and there may be a bit of debate on this. It says, "This is Amos from Glidance. I wanted to bring up a topic that was really brought home for me listening to various podcasts discussing the problems blind folks have when joining a new employer with accessibility of the tools provided by that employer. They discussed the all too familiar issue we have that when we have to use any kind of specialist tools, we fully expect them to be inaccessible. In fact, how often do we make our tool selection in our own lives based on their accessibility? As you know, I started Glidance about 24 months ago and as the CEO, it is important to me that the internal tools we use to run the business are accessible. This is because firstly, I need to use them and naturally I want Glidance to be a place where blind people can join our work."
"There are quite amazing tools out there for start-ups, work management tools, cap table and stock management tools, engineering management tools, collaboration, finance, HR, you name it. It is incredible what is available today for a start-up founder. Yet reflecting back on the past 24 months, considering which are the best tools for my business, guess what is the number one consideration? Will it be accessible? Will I be able to use it? Not the pricing, the features, the user experience, the strength of the company, the suitability and alignment of the provider to our culture. None of these are factors if it is not accessible. And you will not be surprised to hear that almost all of them are sufficiently inaccessible so as to be unusable. Let me repeat that. Almost all the state-of-the-art de facto standard tools used by entrepreneurial businesses today are unusable.
Now, I will be careful how I phrase my next sentence. Yes, most of the tools I ended up picking are plenty accessible compared with those that are not at all. However, even those that are do not pass the bar. What is the bar?"
"I think about it in terms of accessible, usable, and a joy to use. A joy to use financial systems, expenses, HR systems, how can I possibly use that term with such systems? Well, I will tell you that these modern SaaS, which stands for software as a service, systems are a delight to use by sighted people. They look great. They're easy to use. They do a lot of the grunt work for you. They are a significant step forward. Let me share some specific examples. I want to start with the ones that do a good job. Most start-ups use Slack and Google as a standard. Slack have done a great job, but Google Docs, Sheets and pages are not easy enough to use and you will not be surprised that I chose the Microsoft 365 software suite. Why? Because I know Microsoft's commitment to accessibility. It is all accessible and productive. There are no tools than the Microsoft tools that are more accessible today, in my experience. Are they a joy to use? For the most part, they are."
"Although explaining to any of my collaborators why I'm not on Slack and Google is laborious, yet I do it unapologetically. Let's look at work management tools. Examples include monday.com, Asana, Upwork, Smartsheets and such systems. The only solution that is remotely accessible is Asana. Is it usable? Kind of. Is it a delight? Absolutely not. It is laborious. In fact, after your show on Todoist, that was an episode on Living Blindfully, I actually investigated that as a possible replacement. It is less feature rich, but from an accessibility perspective, it is better. Every company has to have a work management tool, and at this point in time, I don't know of a good accessible tool. It is crazy. I think Asana, monday.com, Upwork, and Smartsheets should be berated for their negligence. It is completely unacceptable that their tools are not accessible. All they do is manage lists. If it was Photoshop, I would understand, but really? The leading engineering management tool is Jira by Atlassian. That is the de facto standard in any software engineering outfit. Every engineer knows it. Product managers use it."
"This is how software gets built. I'm sorry to disappoint you. Parts of the traditional apps in Jira are somewhat accessible and usable, not a joy at all. Some such as Product Discovery and Confluence are entirely unusable. A blind engineer joining a company that uses Jira will be severely hindered. I know some engineers such as our own Joel Dodson, who wrote Python scripts to pull data out of Jira just to get by. Again, this is the standard in engineering outfits and there are no excuses here. Atlassian is making it impossible and should be held accountable for making it harder for employers to hire blind people. What about HR management? These are tools that are used by every employee to manage their employee data, benefits, pay stubs and much more. Gusto and a number of other tools have emerged to make life very easy for employers. They take care of payroll, of taxes, of state registrations and all the hard work. They're really great tools. Our company chose Gusto because honestly, it is almost a joy to use. Everything seems to be accessible."
"I was able to set things up, make my choices for the company and open it for my team to use, all great. We created health insurance benefits and opened enrollment, and on the final stage of enrollment, you need to sign a form with your mouse, fail. How am I meant to manage that as a founder and CEO of the company? Do I walk away from Gusto now? I have put in a great deal of effort to onboard onto Gusto, yet I am now stuck with a solution that is not fully accessible. Carta, that's spelled C-A-R-T-A, is a system to manage the company's investors and stock options for employees. It is the standard. Everyone in the tech startup world uses Carta. It is where investors look at their portfolios and where employees review their stock grants and exercise options. Well, Carta appears to be accessible. I get everything set up and I come to issue a new grant of options to an employee to find that the sheet to enter the data is not accessible, bam, fail."
"What do I do? Do I keep looking for a stock management tool that is accessible? I bring this to you, Jonathan, because when we talk about employment and opportunity, a lot of that is about making it easier for employers to employ and when the core systems that employers use are not fit for purpose despite every effort one makes, I'm a blind CEO and founder after all. I feel a sense of despair. In addition, I need to choose systems based on what is right for the business, not based on its accessibility. My responsibility as the CEO is to my stakeholders, my shareholders, employees and customers. It is on me to use systems that serve the needs of the company, yet I have to make choices based on accessibility. I'll leave it here. As you can imagine, I have support calls with everyone involved. I actually think this is low-hanging fruit. It doesn't require legislation. It just requires spotlighting, which is what you do best." Amos, thank you very much for that heartfelt email. I sense the frustration in it and I understand it as well.
At the National Federation of the Blind, we would not necessarily agree with you that legislation isn't necessary. Because if all it took was people pointing out what was required, I guess the fact that you have approached these companies. You're the CEO of a company that's getting quite a bit of attention out there at the moment. You would have been able to move the needle, and I guess many others would have been able to move the needle as well. This email came through before the episode where we interviewed President Riccobono about the Washington seminar. But hopefully you heard that episode where we were discussing among other things, the Websites and Software Applications Accessibility Act. This is something we feel passionate about. You can read all the details about it at nfb.org/washington-seminar. That's nfb.org/washington-seminar, but the synopsis is this, websites are required by law to be accessible. But without implementing regulations most businesses and retailers have little understanding of what accessibility means. The Websites and Software Applications Accessibility Act will define accessibility, provide technical assistance to support accessibility and enforce accessibility regulations.
We would obviously invite you and others who feel passionately about this subject to join our efforts to get this enacted. It is important legislation. We'd like to see it go all the way and become the law of this country, so hopefully you will lend us your advocacy efforts, join the National Federation of the Blind and help us get this one across the line. And there are some other things that you can do as well. At the Center of Excellence in Nonvisual Accessibility that I lead here at the National Federation of the Blind, we do meet regularly with a range of companies. And as we've said in this podcast in the past, what really helps us is quality data about the problems that people are experiencing. So if you are trying to use a piece of software and you're coming across a very specific bug such as a few of those that you mentioned, we would be most grateful for you taking the time to complete our Inaccessibility Tracker Form.
Because when we meet with these companies, if we can point to people who are having very specific problems in the real world, that helps us make the most of those meetings, and it is true. I've certainly found this here in my time at the NFB, that the NFB can unlock doors that it's very difficult for others to unlock. So you can take advantage of that by using the Inaccessibility Tracker Form to report that data to us. Google is a case in point, by the way. We have in the past expressed some frustration about the time that things take with all of these companies, and Google's no exception, but they do meet with us regularly. We have very constructive, fruitful dialogue with Google. And if you are having issues making use of the Google productivity suite, I know that the accessibility team has spent a lot of time on those products such as Google Docs, Google Sheets, et cetera. If you find them inaccessible or just inefficient, we would like more data on precisely what you're doing, what you wish would happen instead of what does happen.
That's really useful data for us, so if we can collate that here at the National Federation of the Blind, we can help everybody. Do feel free to be in touch. Complete that Inaccessibility Tracking Form. Go to nfb.org/cena. That's nfb.org/C-E-N-A, and you'll find a link there. We would be most grateful for that data. You can be sure that we do sit down with Google and go through all of the feedback that we receive. This applies to big tech, but it also applies to some smaller tech companies. As well at the National Federation of the Blind, we are building relationships all the time. And what's happening is when some of these companies reach out to us for advice, none of this is to minimize the points that you are making, which are very well-made. But I do like to try not only to empathize with you, but also to offer some practical solutions about how we can keep moving forward because we firmly believe at the National Federation of the Blind that we're stronger together. Our tech tip today deals with how you can access ChatGPT if you don't have a smartphone.
And I know there will be people listening at the moment, maybe on a smart speaker or a Victor Reader Stream or some device that isn't a smartphone, and you're hearing all this buzz about ChatGPT and perhaps feeling a bit left out. Maybe when you go to an NFB chapter meeting, you meet people who don't have access to a smartphone and would like to be able to find out what all this AI is about. Recently, OpenAI introduced a phone number in the United States that you can call and talk to ChatGPT just by making a regular phone call, and you can text it on WhatsApp. That also means by the way that you can text ChatGPT using your Meta Smart Glasses, which has WhatsApp built in. The magic phone number to do this is 1-800-ChatGPT, which equates to 1-800-242-8748, and if you want to, you can just call this on a landline phone. But I'm going to call it on my handy dandy iPhone that I just happen to have here so let's try this. Call 1-800-242-8478.
Speaker 7: Calling plus 1-800-242-8478.
Speaker 8: Hi, it's ChatGPT. This call may be reviewed for safety. How can I help?
Jonathan Mosen: Hey, it's Jonathan from the National Federation of the Blind. How are you?
Speaker 8: Hi, Jonathan. I'm doing well, thank you. How can I assist you today?
Jonathan Mosen: Well, first of all, I just want to tell you, you are talking on the Access On Podcast.
Speaker 8: Oh, that's fantastic. It's great to be part of the show. What's on your mind today?
Jonathan Mosen: Can you tell me about this phone number? I'm interested in letting people know that if they don't have access to a smartphone or a computer, they can communicate with you this way. What are the features and limitations of using the 800 number to talk to you?
Speaker 8: Absolutely, calling this number allows people to interact with me without needing internet access. They can ask questions, get information, and even have a bit of conversation. However, there are some limitations. For example, sharing images or browsing the web together isn't possible over the phone. It's a more streamlined voice-based interaction.
Jonathan Mosen: Oh, my goodness. Is it something I said? It just hung up on me, but that's the idea. You can talk to ChatGPT. You can text it as well using WhatsApp, the number 1-800-ChatGPT. You only get apparently 15 minutes a month, so maybe that's why it hung up on me. Maybe I've used my 15 minutes already this month, but I don't think so. I typically use the app, but this is a way of giving people a bit of a flavor of what all this AI stuff is about, 1-800-ChatGPT.
That concludes this episode of Access On, the technology podcast of the National Federation of the Blind. To send in a contribution for a future episode, email us, attach an audio clip or just write it down and send it to [email protected]. That's [email protected]. To keep up to date with Access On, follow us on Mastodon. [email protected]. That's [email protected] on Mastodon. To subscribe to an announcements only email list about upcoming episodes, send a blank message to [email protected]. That's [email protected].
To learn more about the National Federation of the Blind, visit our website, nfb.org or phone us, 410-659-9314. That's 410-659-9314 and be sure to check out the Nation's Blind Podcast right from where you heard this podcast.