Equidox AI Demonstration
Watch our cutting-edge fully automated PDF remediation solution in action!
Watch our cutting-edge fully automated PDF remediation solution in action!
[Pat Needles] Hi everybody my name is Pat Needles. I am a Vice President here at the Equidox Software Company. I first wanted to thank each and every one of you for making yourself available today to join us for what proves to be a very exciting and new technology that Equidox has developed and we're super excited to show you how that works and the value that is associated with that. And you can see on this first slide up top our mission statement, which is “enabling PDF accessibility through intelligent automated solutions.” and that's kind of a common theme here you'll see that throughout today's presentation and discussion about the genesis of Equidox. Where we started where we've been and where we're at today. So we're excited to talk to the folks especially we've never ever heard of Equidox. This might be their first time at one of our webinars – but welcome to you and we're looking forward to showing you what we have today. (you can move forward Dan) So what we're going to talk about today as far as the agenda is concerned is, who is Equidox where did we come from, where are we going. As I had mentioned before some super exciting things that we're going to talk about. Why make PDFs accessible. I think most people on this call understand a lot of the reasons why you might want to do that and that sort of thing and so we're also going to talk today about our AI-powered PDF accessibility tool and how that works and the value of that. And of course we'll get into the most important thing here today which is the actual demo of the AI solution. So who is Equidox and where did we come from? Equidox the genesis behind it starts in about 2011 when we were asked by a Canadian federal government to build a software tool that would make PDFs accessible for people with disabilities. And when we started this process and building the tool our first output was PDF to HTML. Soon after that we grew to a PDF-accessible output and from there we have built in a lot of different types of automation into the tool that we will definitely be talking about. We have built over say, over the last three years, some detection tools into our software application that has made PDF remediation much simpler and more efficient. And any of those… any of you folks that have been that have done remediation that is on this call today understands that this is a very painful process. It can be time-consuming and we've taken a lot of that heavy lifting off of the table by building in tools into the software application that allows for fast and easy remediation. And then secondly what we're going to be showing you today is our AI-powered software solution for high-volume templated at scale recurring documents. (okay Dan) so you know we talk about the challenges of PDF remediation and how difficult that can be you know it's very labor intensive. Folks need accessibility skills in order to remediate documents properly. It can be very expensive if you are outsourcing these types of activities to third parties. There is the possibility of human error while remediating documents on challenging Timelines. (Go ahead, Dan) So the Equidox AI solution as I had mentioned earlier is a fully automated PDF solution that really takes the human element out of actual remediation. And so what we're going to show you today is how we import documents into our system, how they run through our AI engine, and how they come out completely tagged and completely accessible. There is no need as I had mentioned for any human intervention for this too. And we are really excited to show you how this works here today. So with that, I'll go ahead and I will turn it over to you Dan, to do the demonstration of the tool. [Dan Tuleta] Okay great, thanks, Pat. So before we get into the actual demonstration of Equidox AI I’m just going to set the table for a little bit more just in case anyone is new to accessibility. I'm just going to talk through a little bit of the process and how it works and also why it's important to do this so just so everyone's aware. Equidox in in cooperation with the NFB conducted a survey of over 250 blind and low vision assistive technology users. So that would primarily be people who are using screen reading technology in order to read digital content on their computer screen. And we have found, after surveying these 250 blind and low-vision users that at least two-thirds of PDFs that they interact with on a regular basis are not accessible. So they cannot interact with those documents if they were to use their assistive technology. Adobe, who owns the PDF format estimates that there are over three trillion trillion PDFs in circulation today. So that means there are about two trillion completely inaccessible documents out there that are floating around in circulation. So that is a staggering amount of documents and many people who aren't blind or have low vision and don't use assistive technology, are probably interacting with PDFs on a regular basis and you don't even really think about it. But those PDFs present a huge barrier to understanding information for people who who rely on their assistive technology to interact with the content. So the main takeaway from from this slide here many people on this call have probably heard the the acronym WCAG, which stands for the Web Content Accessibility Guidelines maybe you've heard of the section 508 standards. We're not going to get into the alphabet soup and start breaking down the differences and the similarities between these different standards and guidelines but the main takeaway from this is that these are the standards that are being used to measure the accessibility of websites and documents and digital content and the number of lawsuits that we're seeing every single year is on a steady rise so the lawsuits are coming in fast and furious around websites and documents and digital accessibility as we move to a further and further digital world world. So organizations have to be aware that they are not exempt from making sure that their content is accessible and you have a responsibility to make sure that you're providing digital content in a way that can be used by everyone. So with that said, as Pat kind of alluded to a few slides back, the the challenge with PDF accessibility for many organizations is just the sheer volume of documents that they produce and distribute. So the the number of pages can make it extremely difficult to manage on a consistent basis. Outsourcing this work is extremely expensive. You also lose control of the deadlines making sure that that outsourced company is getting those documents back to you in an efficient manner and meeting your deadlines. You can lose… you can have concerns over the security and the privacy of these documents if you have anything that needs to remain confidential and also you're putting the faith in these outsource firms that they are going to be delivering high-quality work which they may not be. And also remediating these documents internally for these organizations has a lot of the same challenges but mainly you have to dedicate very large numbers of staff to be able to manage the volume of documents and it becomes a big killer of time for these organizations and the employees that are the ones that are having to do the actual remediation work. So what we have done is we have introduced Equidox AI which can fully automate the accessibility portion of making these documents accessible using artificial intelligence, a combination of machine learning and computer vision. So we are essentially applying training to these artificially intelligent models to teach it what these documents look like, the different structures that are making them up and then it can fully automate the process to ensure that the documents are accessible and being delivered to the customer without any issues of using it with their screen reading technology. So Equidox AI, it all starts with human programming. So I just want to make it clear that Equidox AI is not “auto-tagging.” There are many people who will throw around the term “auto-tag,” meaning that it's automatically creating tags for the PDF and the tags are what are being used by screen readers and assistive technology to interact with the content. Equidox, although it is applying tags in an automated way, it is not auto-tagging. Auto-tagging requires basically hard coding to teach an auto-tagger that this type of font should be a certain type of tag. It is very inaccurate it's very unreliable it will not produce compliant or usable documents. So what we do with Equidox is we start with human work being done up front. We train the AI models based on the structures of that specific PDF template that whatever that template may look like, it could be a bank statement, it could be a utility bill, it could be an explanation of benefits on an insurance form… So we have a combination of our data scientists who are doing the actual AI training, we have our software developers who are implementing this technology, and we of course have accessibility experts on staff who can assist in that training of the AI models because they know exactly how to tag a PDF document. So just to go through a little bit of the process… So Equidox AI is implemented postprocessing. And what I mean by that is you do not have to completely uninstall and recreate a process for building your PDF documents. Many organizations already have a very well-established process in place to create and deliver these documents to their customers. We are are placing ourselves into that workflow seamlessly without having to completely rebuild that process from the ground up which could be a very time-consuming and costly endeavor to do so. The first step with Equidox AI is, you would provide sample documents to us in order for us to analyze the documents. And then we would train the AI models based on how we analyze those documents. So we teach the AI models all of the different elements that are that are accounted for in your template. We apply that training and then we would integrate the workflow so we would figure out how do we want to deliver these documents to the customer. Is this something where they are downloading a PDF from one of your portals? Maybe it's their bank statement, for example, or is this something where you would like to manage it internally, where perhaps you have a bulk import of documents every single month that need to be made accessible before they go and get posted online? Once that is that decision is made as to how we're going to integrate the workflow, the process would begin where Equidox would apply the tag structure to these documents. The documents are then exported from the system and then they are delivered through whatever mechanism we've determined. Whether it's directly to a specific customer of yours, or if it's to some sort of data repository so that these documents can be uploaded to your internet or website or wherever it needs to go. So just a a quick rundown of how we analyze PDF documents we don't need to get too technical with this but basically these color dots that we're seeing on the screen here these color dots just represent the AI's analysis of different variations of documents. So they are grouped together in different areas by the AI based on different anchors that it's seeing. So for example these documents all contain pie charts so that is what would group them together in a specific cluster, and it would trigger the AI to look for certain variables and certain elements on this PDF that are consistent with the training that we have provided. Here's another example where you can see that a number of documents, in this case, are sorted into different segments based on the number of columns. So you have certain pages that might have two-column layouts. Other pages might have three or four or a single layer or single column layout per page. So it's always based on the template and the training that we provide from that initial analysis of the documents that you've given to us so that we can teach the model exactly how to tag the various elements. One final thing to call out is just the ability to even tag tables. So if you think of like an easy example to think of would be like your credit card bill it's a very static document could be the same document template can be provided to hundreds of thousands maybe even millions of customers for a credit card company. And that document is being produced on a monthly basis. Certain customers of that credit card company might have a single charge for the month on that credit card and their statement might be a single page with a single line item for a single charge in that table. Another customer might have 500 charges on that credit card for the month and that statement, although it comes from the same template, will be very different. It might be 10 pages long and have page after page of tables. So we are able to actually find these variances and we can account for them to make sure that everything is being tagged accurately. So it's not just a static layout on every page there can be variables across these templates and we're able to train the AI models to identify these variables and account for them accordingly. So the the training of the AI model, just so everyone is aware, we do define the different elements. So if you've never tagged a PDF document before, PDF documents will contain primarily text but there has to be a structure applied to the the text on the page. For example, there are headings that are used for navigation purposes. Tables have specific layouts that need to be tagged in a certain way so that assistive technology can interact with them. There could be lists. The reading order is a very important aspect as well. If a page has three columns a screen reader needs to be told to read down the left-hand column first, then the middle column, then the right-hand column. If it's not trained on how to do that, the screen reader could easily just read clear across the top line of all three columns rendering the entire page as completely useless. So these are the types of training that we that we put into the the models to teach it for all of these specific variables for the specific template. And then of course we have to test the model. So this is not just a one-time we teach it what this document looks like and we just hope for the best. We of course test these models and continue to train them throughout. And so our goal is always to both pass the automated checkers as well as making sure that these documents are fully usable for the end user who's going to be interacting with them. So we do apply both manual checks as well using screen readers to sort of replicate how a blind user might be interacting with that document. So these are very very important aspects of PDF accessibility not just auto-tagging and hoping that it passes an automated checker. It really only matters if the end user that's receiving the document is able to interact with it and that's what really matters most to us. And also, I alluded to this earlier, but there are different methods of deploying Equidox AI technology. So we do have we do have an interface that we have built which I will be demonstrating here in just a moment. So we've built an interface that a customer would be able to run this technology themselves if they prefer. We can also embed this technology through through APIs. So like I said before, we can make sure that this is fitting into your existing workflow and you're able to apply the tags and deliver an accessible document in a fully seamless transition where no human is required at all. And then we can also operate this as a managed service. So we can run the entire process and technology for you and deliver those documents to you, depending on the use case, depending on security requirements, anything. Every customer may have some different requirements and we're here to work with you and figure out the best method of delivering these documents in an accessible way. Okay so what I'm going to do now is actually transition out of this slide deck and we will get into the actual demonstration of Equidox AI. So if I just go to this other tab here I have in my browser, this is the interface that we have built just to kind of demonstrate the technology to make it easier to understand what's actually happening. So right now I have a completely blank slate I have nothing really to work with as we can see so what I'm going to do first is I'm going to go to the Upload Documents tab. When I arrive at this tab here I can then just open up the folders on my hard drive and in this case I'll just drag a .zip file that contains some bank statements that we have scrubbed. See these are just like sample bank statements and we will upload that .zip file containing the financial statements. So once these documents upload, I can then go to the Run Batch tab here on the left-hand side. I will then choose the model that I want to apply. In this case, here we've only got three models currently uploaded to my little interface here for demo purposes, so what I'm going to do is I'm going to select the statement model. I will then choose the .zip radio button here and I will select that .zip folder that I have just uploaded just a second ago. Once I've selected the .zip file and I've selected my appropriate model, I just hit Run Batch and this will kick off a process where Equidox is currently ingesting the documents. It is looking at them kind of page by page, one by one and it is taking all of that model training that we have done ahead of time and it is automatically tagging these PDFs. So we'll see these green lights start to light up here as we work through the process. So you can see one of them is already complete, and we will see more of these green lights kind of lighting up in just a few seconds. Now keep in mind that this is just a demo environment of this technology. Equidox AI can be scaled to meet much higher demands and throughput. Of course, certain organizations might require literally hundreds of thousands of documents every single day to be remediated. We can just dedicate additional resources to it in order to power it to meet those types of demands for speed. While these documents are going through the process, I'm just going to open up one of them here just so that you can see what these documents look like. So if I open up like randomly statement number four here, what I can see is that this is a sort of generic-looking bank statement. We have images up here, we have text, we have headings, we have tables. But this document is completely untagged in its current form. If a screen reader was to interact with this document it would tell them that this is a blank PDF. So they would not be able to interact with this. They would have no idea what their monthly transactions were. They wouldn't know what their balance due was. Obviously, that's a huge problem because you want to… some people really need to make sure that they are being charged appropriately for all of their transactions for the month and they, of course, need to know how much they need to pay for their credit card bill. So we need to make sure that these documents are being tagged and delivered in an accessible way. So with that said while I was talking there the whole process finished. I can see the status is now done. So all of those documents finished processing. I can look at the individual PDF files. So if I were to let's say just download one of them I will download this and save it to my desktop. And if I were to open this up in Adobe Acrobat, I will see that this version of the document is now fully tagged. So if you're familiar with tag structures, you know how complicated it is to manually set up these types of tag trees in Adobe Acrobat. If you're doing it manually it's a very time-consuming process. It requires a lot of technical expertise and even if you are the best Adobe Acrobat user in the world it is still very very time-consuming. So at the scale of things like statements, where there could be literally hundreds of thousands of them produced every single day by a single organization, it is completely unsustainable to expect human remediators to go through the process of tagging these elements manually. But now if I look at this version of the PDF, I have the logo has been described. We've trained it to apply this company logo alt description to the logo whenever it sees this particular image. We have text identified so the entire document is tagged. As we can see here, we have text identified, so the bank name and the customer name. We have our figure tag which is just the image, we have our heading level one. As we tab through we can see we have an H2 for the account summary, the table which is nested below it has been properly tagged. So you can see all of the rows of the table are properly tagged. We jump down to another heading level two for the deposits and credits, another table that is properly tagged, another heading, another table… All of these different elements have been accurately accounted for because of the initial training that we have put into this document template. So at the scale of a company that is mass producing these types of documents, we can now fully automate that process so none of your staff has to spend countless hours of their week remediating PDFs. This type of process can be applied to other templates. So it's not just for bank statements. It could be explanations of benefits, it could be bills, like utility bills or medical bills, or medical statements and reports and test results. There's an endless amount of of use cases for Equidox AI. It really just comes down to the customer and the types of documents that they're producing. But we would certainly love to chat with you based on your specific use case and your specific template to see if Equidox AI would be a good fit for your organization. So closing out of this document now I can also download all of these documents in a batch as well. So if I go to my Batches and Jobs Tab here, this is the batch that I just ran. So I would be able to extract all of these documents in a .zip file as well. So I'm able to download them as a .zip just like I delivered it to Equidox. But keep in mind you don't always have to run this process through this through this interface. We can run this process for you as a managed service, or the technology that's kind of going on behind the scenes can be seamlessly placed within your existing workflow so that we are not disrupting a longstanding process that you have for generating and delivering these PDF documents. Okay so just as a quick summary. Equidox is a leader in PDF remediation. This is all that Equidox does. We are entirely focused on PDF remediation. So we can solve your PDF remediation challenges whether it's a large or a small project. Many of you are here and this may be your first introduction to Equidox. Keep in mind that that is only part of what we offer. So the Equidox AI solution is our newest technology. But we also have Equidox Software that is meant for more unique one-of-a-kind ad hoc documents. So there are many organizations that are producing you know marketing flyers or just documents that are meant for kind of one-off use. They might not be producing them en mass. So Equidox software is designed to dramatically speed up and make more efficient the process of going through and tagging those documents that have completely unique elements where we can't necessarily train artificial intelligence to fully automate that process. We have tools built into that Equidox technology that are powered by artificial intelligence to speed things up, but there are certain examples of documents where you do need some sort of human intervention to analyze the PDF for context and be able to properly remediate it. So if you're interested in those types of use cases you can also feel free to reach out to us after the fact. We would love to talk to you about both or either or use case that you have in mind for your organization. And of course, we did just see the Equidox AI demonstration there. So you have a general idea of how that technology works. We will be sharing this slide deck out to everyone on the call after after the fact. So there are some relevant articles here if you are new to PDF accessibility or digital accessibility. If anything of anything presented today was confusing we will certainly give you some resources of some relevant articles and we do have a really large library of information on our website as well. So feel free to interact with these articles and check out some of our past webinars if you're interested in seeing our other technology. But also please feel free to reach out to us for a more personalized demonstration and discussion around your use cases. So with that said I'm going to pass it back over to Pat. We just have a minute or so left to to kind of wrap things up but thank you everyone for for attending today. [Pat Needles] Thank you, Dan. Great job great presentation. I hope that all of the folks here on the call today saw the great value in the Equidox AI tool. And we'd love to hear from anybody who feels as though there might be a proper use case not only for our AI solution but as Dan said our PDF ad hoc remediation software tool. And you could reach us at EquidoxSales@equidox.co, of course, you can always reach out to us by phone you see the number there, or if you would just like general information feel free just to visit us at www.Equidox.co. And that concludes today's presentation. For more information about how Equidox Software Company can help you with PDF accessibility, email us at EquidoxSales@Equidox.co or give us a call at 216-529-3030, or visit our website at www.equidox.co.
Are you curious about automated PDF accessibility? Now you can witness Equidox AI in action!
Free 30-minute webinar where we’ll walk you through the process and offer a live demonstration using a variety of sample documents.
During the webinar, we’ll show the before-and-after results of Equidox AI’s automated remediation process, where our AI-powered models generate accurate tag trees at the click of a button.
Don’t miss the opportunity to witness this cutting-edge technology in action!
Speak with an expert to learn how Equidox solutions make PDF accessibility easy.