Hosts Eamon McErlean and Joe Devon interview Eugene Woo, CEO of Venngage, a company specializing in design tools that create accessible infographics. The conversation focuses on Woo's background as an engineer and entrepreneur, how he started Venngage, the company's commitment to accessibility, and the impact of AI on Venngage’s design features and the broader industry.
OUTLINE:
00:00 Opening Teaser
00:53 Introduction
01:24 Eugene's Background That Led To His Current Role
03:18 What Inspired The Idea for Venngage
06:23 Lessons Learned Being An Entrepreneur of Start-Ups
10:40 How Do You Maintain The Start-Up Culture As A Company Expands
15:04 Venngage's Decision To Focus On Accessibility
19:39 How Venngage Incorporated Accessibility Into Their Product
21:52 Favorite AI Models and Platforms
25:11 Demonstration of Creating An Infographic About Accessibility
36:33 What Are The Testing Processes For Accessibility?
40:08 The Future of Venngage
42:36 Are You Using AI To Build Venngage?
44:03 Have Your Customers' Expectations Changed Since The Use of AI?
46:50 Will AI Democratize Access To Technology or Make It Worse?
48:49 Will AI-based Tools Automatically Make Documents Accessible?
52:33 Wrap Up
--
EPISODE LINKS:
Venngage
https://venngage.com
Startup Weekend
https://en.wikipedia.org/wiki/Startup_Weekend_(organization)
ServiceNow Accessibility Statement
https://www.servicenow.com/accessibility-statement.html
Eugene Woo
https://www.linkedin.com/in/yjwoo/
Venngage on LinkedIn
https://www.linkedin.com/company/venngage/
- So this is our sort of default AI designer. I can put in, I create an infographic about AI and accessibility. It's not an image generation. It actually generates a, you know, a multilayer design. You can see everything is a layer. You can generate images, you can generate, you can improve the text, and you can rewrite any of these if you want. And then from the accessory point of view, it literally goes through the whole list of what WCAG requirements for an accessible document. So for example, you know, color contrast paths, which is great. All text did not pass. So all of these icons, you either need to add an alt text or you can just mark them as decorative. So you can see it's very easy for you to kind of like modify things.
- Welcome to episode 13 of Accessibility and Gen AI, the podcast where we talk to people who are shaping the world of accessibility and artificial intelligence. I'm Joe Devon, joined by my co-host, Eamon McErlean and today we're speaking with Eugene Woo, who is the CEO of Venngage, a company that's dedicated to making accessible design tools. Eugene, welcome to the pod.
- Thank you for having me here, Joe and Eamon.
- Absolutely. So why don't we start with you sharing a bit about your background. From what I understand, you were born in Malaysia, moved to Canada. By the way, I'm Canadian from Montreal. And then I saw that you pursued a Master's of Science in planning. Can you explain what that is and what drew you to that field?
- Sure, so I actually went to the States first. So I lived about my 10 years in the States. It's where I got my undergraduate. And that's why I left Malaysia. It was for my undergraduate. I went to university there. I did study computer engineering, and electrical engineering. So I do have a software background, but then I think somewhere mid-career I thought I would try to do something different. And planning was this multidisciplinary kind of a degree, which is all about solving problems at a much higher scale. But unfortunately, you know, what happened there was, I think I got introduced into politics and I was like, oh, planners don't actually make decisions, the politicians do. So I, you know, I don't really want to be a bureaucrat. And so I went back. After that degree, I actually went back into software and yeah, so that.
- Smart.
- That was a very long story. I think the problems were really interesting. I was really interested in solving like sort of citywide. And so I was, it was a sort of planning program. And I love, you know, I love being in a city. I love solving like, traffic, environmental, bigger scale problems. But I realized, you know, as I mentioned that, okay, planners don't actually solve these problems. At least their suggestions aren't actually the ones that get implemented. And so I, you know, so I kind of did this kind of reverse turn and went back to software.
- Eugene, we haven't met in person, so pleasure to meet you. Thank you for joining us this morning today. So solving problems and not wanting to be a politician, like how did that transition into the creation of Venngage?
- Yeah, so Venngage, this story really, it's about, it's almost 15 years now, I think. It was from a problem that I had. I essentially was trying to create these. So infographics have been around for a long time, but I think sometime around the early, you know, 2000s, it became really popular and everybody was making it on the internet. Not everybody, but people were beginning to make them and put them on the internet. But it was mainly designers, like professional designers who were doing them on Adobe Illustrators and Photoshops and those kind of tools. And I'm not a designer, I'm like an engineer. So the first thing I wanted to do when I saw this was like, this looks great. I wanna create something like this. I wonder if I can just code it and do it. And the first prototype of Venngage was actually called Visualize Me. And what it did was it would take your LinkedIn profile through an API and it would visualize your, like your resume, essentially your experiences. You had like a bar showing how many years of experience of certain skills you had. You know, like a timeline to build up this really visual timeline. And I think it had some pie charts there as well. I don't really remember. And back in the old days that I think LinkedIn gave you account of your, now it just gives you a kind of like 500 and above, but it used to give you account, so we would also visualize your follower account or your, sorry, your connection account and all that. So that was the genesis of Venngage. It was like, I wanted to be a designer, but I'm not a designer, but an engineer. And then that eventually became Venngage, so.
- And did like, when you were doing that, did you believe that this would turn into like an ongoing company, an ongoing business?
- No.
- No? No, no. So when I did that, it was actually part of a hackathon. So there's this hackathon called Startup Weekend. I don't think it exists anymore, but it was kind of popular, like, I don't know, 15, maybe even 10 years ago was still popular. It was, so I pitched this idea. I was like, hey, I'm gonna visualize your LinkedIn, your resume on your LinkedIn with this app that I have, this web app, and it was gonna be a weekend thing that kind of turned into a little bit more than a weekend, that eventually turned into like a company and a product.
- Wow, it's interesting. I've been to Startup Weekend . Was a judge on it once. Helped out a team that they were coming up with an interesting angle for sort of displacing LinkedIn or whatever you wanna call it, a disruptor, where you, any team that you've worked with, essentially you can bring them into the system by sort of nominating them. Like you needed to have that invite in order to get part of the team. And this proves that you're part of the team. And his idea was to allow folks to hire from that team, which is pretty cool. But I saw that you actually turned this into an exit for yourself and you've been an entrepreneur for a while, haven't you? With multiple startups that have gone sometimes great, sometimes not as great. Can you share what that exit looked like and what you've learned from some of those ups and downs of being an entrepreneur?
- Sure. So prior to that infographic company, Visualize Me, I had like two other companies, startups you could say. One was just a development shop, you know, services. So I was an app developer before that. I was developing apps for, you know, doing a boom, I would say the boom of iOS store and all of that. It was always great. So I was developing apps, but as most agencies and service providers know, you always wanna do a product. So when Visualize Me happened, I was actually still kind of working on the services company. And then I would say six months in Switched, went all in into Visualize Me, which was the infographic resume company that got bought very early on. We were like barely a year into it. In fact, I was starting Venngage. So the genesis of Venngage was like I needed to monetize this infographic generator that I had, but Visualize Me was a consumer tool. All it did was just this one. It had like one little button that you pressed on it and it gave you the infographic that you could edit it, but very limited. And I basically didn't know how to monetize it. So I created, so I said, hey, I basically, why don't we just do, let's make this drag and drop. At the same time as Canva, and a lot of these other tools, like this is 2012. So we all kind of had like the same idea at the same time, they just executed a lot better than us and had a lot more money than us. But, so we, around the same time, you know, Venngage got started, someone, Parchment bought us and I went to work for Parchment for about a year and a half. I didn't last very long there, because at the end of the day I wanted to be an entrepreneur and I wanted to go back to Venngage, so. It was a great experience though. I think it was a very positive experience. I, you know, obviously, you know, there was some monetary, you know, up, you know, there was that windfall there. And then there was also, I also learned from the CEO of Parchment, this guy called, he's also a, he's a very successful, he was a former founder and CEO of Blackboard, the educational like LMS.
- Yeah.
- So he, yeah, his name is Matthew Pittinsky. So I learned a lot from him and how he ran his companies, very different from how I run my company. I'm like bootstrap and all that. He was raising a lot of money, and you know, buying and acquiring a lot of companies. So it was very fascinating for me to see how you grow companies that way.
- Totally different game.
- But that was my journey. Yeah, it was a completely different game, yeah. And I think part of what I wanted to do, it did sort of make me think about how I wanted to run my own company. And I think, you know, at the end of the day, I wanted to run the company at a much smaller scale. I wanted to be independent. I didn't wanna have investors. I didn't raise any money. You know, I got some grants here and there. I got some Canadian grants. They're not grants, they're loans. I had to pay them back, so.
- Not as good. Yeah, it's not as good, yeah. I realized like I had to pay with interest. But yeah, but it was a good experience overall for sure.
- Yeah, it's funny like, we've all been part of startups. I think the biggest lesson for me doing the startup is to realize how much I didn't know and how much I had to learn. Like, when you do a startup, it's all encompassing. It really, really is. And you gotta understand all aspects of the business. And this is a question, Eugene, we ask to many of our guests, and this is kind of a twofold question. First, like, what were your biggest lessons through those startups, lessons learned? And then two, to me, all about culture and trying to keep that positive culture. As a company grows, how do you instill and keep and retain that hungry kinda startup culture?
- I think initially, as you said, like I didn't really know anything about running a company. So initially I think the biggest learning was, so I knew other build products, 'cause I built a lot of products for other people and for myself. And so that wasn't, you know, so the learning for me was really how do you hire. Aa lot of it was people problem. Learning how to deal with people, how to manage people. How do you get a, you know, group of people to work together on a mission and a goal. That was probably my biggest learning and the most surprising thing. 'Cause I thought, oh, startups will all be like product and marketing. I mean, I learned a lot about product marketing too, don't get me wrong. But I think my biggest learning was like, oh wow, dealing with people is like really difficult. And that's where all my stress and all of that is coming from. So I don't know if that's surprising to you, but I think it is surprising to most. And people come from, like, the engineering side of things is like, oh my gosh, the biggest thing you actually have to figure out is people. It's not the, you know, yes, the product too, but it's people.
- Yep.
- I would say that was the biggest learning for me was that, was managing people and trying to get people to do something together. From the culture point of view, that has been a journey. So it has changed quite a fair bit. I think initially I thought, you know, if you think about like maybe pre COVID where tech was really hot. It was all about like this very superficial culture of having very nice things in the office, and which we did too. Like, we had a very nice office. We had, you know, free lunch, free meals. Like we, even breakfast, like we had like a grocery cart. You just add lists on the, like you could buy groceries. If you worked for Venngage, back in the day, you would just add it and then there'll be a Costco order that goes out. It'll come in. And then you would have, let's say you wanted to eat, I don't know, granolas with some special yogurt. It would be delivered and then you would eat it.
- Nice.
- So we had all of that, and what I realized is that all of that is just very superficial and it doesn't really form the culture that you want. And at the end of the day, you know, culture. It's really, it's really all about like core values and, you know, finding people who have this, who have similar core values. And I think for me and just like many other entrepreneurs, is that value of continuous improvement and curiosity. Can I find someone who's just very curious, curious in the sense that they want to improve themselves. They wanna like, figure out new things. For example, AI. Like, you know, there will be people who are just not curious about AI, to which I will find surprising, but I think a normal person, they'll probably be normal, right? You know, we're in a bubble. I think if people are in a bubble, we think everyone is like super curious. They wanna try all the new things. But I think the reality is that most people probably aren't. Like, I mean, I feel like most people aren't like this, right? So part of that, so part of my culture was finding people who are like that, who are curious, who wanna learn, who wanna improve, who aren't just satisfied by just sitting and doing the same thing over and over again, getting a paycheck. So that is how I kind of define culture now, which is very, a lot more difficult to interview for.
- No, great response. It is a great response. It's hard to create and it's even harder to maintain as you grow. It really is. But it's pivotal. It's foundational to the overall success. A hundred percent agree.
- Yeah, and it's funny talking about all of those, the couple of things you said was interesting about the culture of tech and Silicon Valley, where you could have your breakfast, your massage, and anything that you wanted. And then you also mentioned about how the normie maybe is not that into AI. And I think from what my experience, when I'm speaking to people that are not in tech, I think there's a lot of fear around AI, because instead of getting the massages and the lunches, there's gonna be an AI employee that's taking over for all of that. So that's, I guess that's a big part of the concern that a lot of folks have. But speaking of culture, you did have a shift at a certain point in your company, in the culture, and a shift in what you were doing. Can you speak to how Venngage sort of started out as a design tool, really, and then moved toward being more accessible first?
- Sure, so I wouldn't call it a big, I wouldn't call it a big shift. It was more of a, hey, you know, we made like a key decision that sort of changed how we worked and obviously changed our product as well. And that is the move towards making our designs accessible. And there's a, you know, and you know, every time I say that, what I really mean is, is that there's, you know, there's the tool itself being accessible. So, you know, that means, you know, someone with, you know, with a disability can use it using a screen reader or something else. So that's one part of it and that's ongoing. And then there's this other part which is I think the thing that makes it, that makes Venngage very powerful is that you can export a design in a PDF that is accessible. So when you share it, it now becomes, you know, becomes a multiplier. So it becomes, it's not just a tool is accessible, whatever you make from it is also accessible and it's also differentiated, because most the other design companies and tools like Canva and Adobe, like they don't do a good job with that other part, the latter part of making the actual PDF, the actual thing that you export and share accessible. And that should happen, you know, five, like, probably like five years now. Like three, four, five years ago where we were in the process of building out our, we were in the process of refactoring our products. So the product by then is, you know, it's almost 10 years now. And by then almost 10 years and it would hit a lot of technical debt when you build. So when you build technical product, technical debt, which is like all the crud and all these weird stuff that just gets added to it over time, just becomes very difficult to add new things. And it becomes just, you know, unbearable at some point that you're like, oh, I'm just gonna rewrite this whole platform, which we did, you know. I usually don't advise people to do a complete rewrite because it's usually not the right move. But we did it. And when we did it, of the biggest things that was missing from our tool was that accessibility part of it. We knew our platform was not accessible. We knew that the exports were not accessible because we've had a lot of people ask us. And the answer we would give, we'd be like, nope, it's not. And, you know. And so when we had the opportunity and the reason, you know, we didn't do it is because we didn't know about accessibility. I mean, we built it like, you know, 15 years ago. We didn't build it to be accessible at all. And it was a lot of effort to, you know, I would say back then it was almost impossible to make it accessible because of all of the technical depth that we had. So when we had the opportunity to refactor, we made it one of the pillars. Actually it was one of the main pillars was like, hey, let's make it accessible. Let's make the export accessible. And then we work backwards and go, like, if that was the case, how would we architect, you know, the way this PDF is? How do we architect the widgets and the background, so that when we export it or actually, and allow people to, you know, to do things like change the reading order. You know, do all the things that you need for it to be digitally accessible. How can we do that? And that's been a great process. 'Cause we, you know, we invented a few things that was really cool. Having having the color contrast in the color picker. I think we're the first design tool that has it. Like Figma has it now, but they only did it like, I think six months ago or a year ago. We had it like, I don't know, three years ago where, you know, it automatically checks their color contrast as you're picking the color. You know, you're not doing it after the fact where you, you know, which is how people do it now. They pick a color, then they go to another tool, you know, and then they put in the hex codes and check. So we kind of invented some stuff that was pretty cool because we had that focus on accessibility.
- Yeah, yeah. No, it's a great call out and great commitment. But to your point, it's not just, it's not just creating accessible experiences, accessible graphics, it's whatever tools that are utilized to make those accessible as well. We're doing the same thing at ServiceNow. We're ensuring that, yes, the platform's accessible, but whatever tools our customers use, outta the box tools that we provide, those tools are also accessible. How did you, you said you committed to that and made sure you put time in to do a revamp. What did that commitment look like? Like was it all inclusive of leveling up skill sets, working more with individuals with disabilities? Like what was that holistic approach to accessibility when you decided to make that all work?
- Yeah, it was all of that. So we, because we didn't know a lot about accessibility, we basically worked with one of the, we worked with one of the big, we hired one of these big accessory service providers. They were bought out by Level Access, but it was a local company in Toronto called Essentials, Essentials Accessibility, I believe, they were called. They were eventually, you know, bought by Level Access. So we worked with them and through that, and then we also worked with a group of PDF remediators, because at the end of the day, that's the problem we're solving. It is, you know, it is PDF remediation, which is this process of taking a PDF and making it accessible. Whereas our platform, you would export and the PDF would not require remediation. So we also work with that group of people and then through both groups, you know, we have, you know, we were introduced to, you know, people with disabilities, people using different, you know, different types of technologies to kind of access our tool and to read the PDFs. So yeah, so I'd say we went through the whole cycle of like actually hiring someone that knew a lot to guide us to working with people who are actually the end users.
- Well, making a PDF is one of the hardest things. Making a PDF accessible is one of the hardest things you can do in accessibility. So you definitely started, you decided, all right, I'm gonna jump off the deep end over here and really work on that. And then after you had this whole rebuild, all of a sudden AI comes out and not that it, I mean, it's been around for decades, but it's ChatGPT moment came around, so you had to do another rebuild. And before we get to the AI portion of our conversation, we definitely wanna see a demo, but just as a preview, I'm curious, do you play with a lot of the different AI models? What's your favorite platform harness or model, or all of the above?
- Sure, yeah, I do play. I don't, I mean, there's only a limited amount of time. As you know, there's so many models. So I definitely would like to play with all the models, but there's just so many of them. In fact, in your notes, you were mentioning this Chinese image model, which I didn't know about it. And I was like, wow. I thought I knew all the Chinese models, 'cause I've looked at them extensively and I didn't know the one that you mentioned. And I was like, oh wow, I should check it out. So the main ones that we play with are the, you know, obviously there's this ChatGPT, Gemini, especially Nano Banana, which is the image generator. Nano Banana is just so far ahead of all the other image models right now. I mean, so as somebody, so I would say I spend a fair amount of time with image models because we're a design company. Nano Banana definitely feels like it's a generation ahead of the rest.
- Yeah.
- Not in just the images, more so in the understanding part. It really, the prompt understanding is just phenomenal. It's just phenomenal.
- It's a real multimodal, which is where I think.
- Yeah.
- That Journey probably is gonna struggle. Like I'm worried about them. 'Cause you like to see like an independent model.
- Yeah.
- Or company really succeed here. I don't know what's gonna happen with them, but you play with them also?
- Yeah, I think, you know, if you look at the, I mean and I completely agree with you, like, like say pre, you know, before Nano Banana, you can get away with just training with images, but now you need the LLM as well because you need all that understanding, right? Like if someone's, for example, someone wants to create, let's say, a diagram about, you know, AI and accessibility while you need all the LLM information. So if you're just training images, you're gonna get gibberish. You're gonna have a beautiful picture with gibberish. But if you have multimodal which is Google and ChatGPT, like you'll get like a really nice visual and really nice content with that visual both, which you need both. You need the context and you need the visuals as well. So unfortunately, I think unless Mid Journey and the smaller models kind of invest in, in also, you know, adding LLM's, becoming multimodal, which is a massive investment as we know, I mean, I don't think they're in the position to do that.
- They don't have the money to do that.
- They would have to raise billions of dollars, you know, which I don't know if they're gonna do, I don't think they were gonna go do that. They will fall behind, unfortunately, they will fall behind. So yeah, so Nano Banana and I've been using Gemini a lot more. So I used to be more of a Claude and a ChatGPT person. But I've been using Gemini since Gemini Trio came out, which is only like a week or two ago, I wanna say. I dunno how. Like very recently I find myself using Gemini a lot more because I find it to be really good. So maybe this is when I, you know, Gemini is sort of like, you know, kind of surpassing ChatGPT almost, and maybe even Claude. You know, I don't know, but I've been using Gemini a lot and very, very satisfied with what I'm getting.
- Do we wanna dive in to the demo? Do you want to give us a look Eugene?
- Sure.
- What it looks like? So I guess I could show the AI features, I could show the accessibility features. I dunno what we wanna do.
- Maybe a little bit of both if possible.
- A little bit of both, sure, yeah. So this is our sort of default AI designer. I can put in like create an infographic about AI and accessibility. So I'll show you a little bit of generator. The generator, it's not an image generation. It actually generates a, you know, a multi-layer design. We are building one that's purely image generation power, but that's not productized yet, so. So I'll show you a little bit of the AI features and then I will show you the accessibility features.
- [Eamon] Perfect.
- [Eugene] So yeah, so it does, it does do image, it does do image generation, so it will try and, yeah, so it will try and generate images that are, these look like they're image generation. So if I click on customize, it will bring me to the editor. So it does generate something that has, so this is an image, like everything is a different, you can see everything is a layer, everything is a different layer. Yeah, these are all different layers where all the AI tools, sorry. So, we've added this design assistant. So we can do, I mean we can do several things right now. You can sort of rewrite this if you want at a very high level. You can, I mean, you can do things like, I want it to, I want it to have a different color scheme.
- And everything you're showing is in production now?
- Yeah, yeah. This is production, this is on Venngage.com. I'm not showing.
- [Eamon] Love it.
- I'm not showing. So you can kind of then kind of shuffle your colors. That actually made it worse, I think. Yeah, I don't think this is necessarily any better. You can rewrite, you can like actually change the whole thing if you wanted to, but I'm not gonna do it. It takes a while to do that. So we're building out the AI designer that, you know, and adding more and more tools. So if, you know, if you know anything about AI agents, it's really just a bunch of tools that you add to it and an agent just picks what tool it can use to do certain things. So you can like click on this and say like, yeah, change the image to, you know, replace the image with something else, you know, for example. So yeah. So that is, so that's the AI tools. Everything else I would say is what you would expect. You can generate images, you can generate, you can improve the text and you can rewrite any of these if you want. You know, you can rewrite any of the text, which I think it's probably, obviously you can translate things as well. And then from the accessory point of view. Yep. From the accessory point of view, what our accessory checker does is that you can see it's a lot of failures here. The AI at the moment is not that good at when it generates it, it's not very good at generating very accessible designs, but they're fairly easy to fix. So what our assess checker does is it goes, it literally goes through the whole list of WCAG requirements for an accessible document. So for example, you know, there was a bunch of small texts. So these are, you know, it claims it's too small, like there is no actual, there is no actual like number that says, but you can make it a little bigger. Like that would probably pass it. Oh, and I was gonna show you the color picker. So you can see this is the color picker here. And we have, so if you pick something that's fairly, that's a little lighter, let's say this one. Okay, that one's a very obvious one that you can barely see it. It will, so if you pick that for whatever reason, it'll tell you it's slow contrast.
- [Eamon] Yeah.
- [Joe] Yep.
- [Eugene] And you know, if you wanna know more, you can do this. You can go into, you can visualize, you can also simulate it if you wanted to. What I mean by this is you can also simulate it with these different types of color, basically colorblind versions of different, so you can do that. But lemme go back to this. Yeah, so this is the color checker. It also works on like, things like this too, or things that you can change colors on. You cannot change colors on the right, so. Oh, and you can see this alt text here. So you can see, you can put in alt text, or so you can just mark them as decorative. So back to the accessory checker, as you can see, you know, color contrast pass. Well, which is great. Alt text did not pass. So all of these icons, you either need to add an alt text or you need to just say that they're all, you know. You can see it's very easy for you to kind of like, modify things. And then the most important thing, I think would be the headings. So you can see it gives you a high level order of your, basically your visual hierarchy or your headings. So an accessible document has a very, it has a hierarchy that makes sense. So this should not be a P, for example, this should be, you know, an H1, and this should be an H, you know? Maybe this is a P, yeah. That's an H2, so that's fine. And then the rest are probably fine. It's H you know, maybe this one you wanna say, oh, I'm gonna make this an H2, for example. Maybe, I mean, in this case maybe not. It's just a number. And everything else is A, is a P would be fine. So that's hitting, so it actually goes through this whole check. And from what I, and you can do it all on the editor. You're not exporting this to an Adobe tool, which is what remediators do right now. They would like to do all of this in an Adobe, basically you would, you're doing this as you're editing this. You're like, let's say you added something. You're like, oh, I want to add a, you know, I wanna add a heading three. Well, if you added something as a header three, it's done. It's actually, it actually is a heading three. And then you were just, you know, add it and you would, you know, let's say, let's say we added it here and then we actually make it so that it's, you can see it's, you cannot read it. And then you have another header here. Let's say you type something in here, right? Like 10 tips for AI accessibility, for example. Yeah, so now you have, this will be tagged properly. I think the last thing I wanna show you is the tag order. So one of the other aspects of readability is that you are, you know, the, you are telling the screen reader this is the order you're reading it. And for a lot of documents, the reading order is sort of like the Z level or where it's placed, which is all kind of all over the place, which it is right now, because this was generated by AI. You can see this is the reading order right now, which is completely like messed up. But the cool thing with this is that you can click on it and you can say, this is how I want it to be. You know, you can just drag things. You can drag the reading order. And then you can say, I wanna read that first, and then I read this, and then I wanna read, you know. I don't think, you know, number one, I wanted to read number one and so forth and so forth. And I wanted to read the, that's the part number one. So you can see now it's gonna read this and then it's gonna read this and then it's gonna read this, sorry, the number one first, and then it's gonna read this. So you can see how easy it's. We've made PDF, like we're not, we're kind of remediating it. We've made it like super easy drag and drop, which if you try and do this in like an Adobe, so you can do all of this in like the Adobe Acrobat tool, pro tool. You basically need to take like, you know, like several modules of training and a lot of, you know.
- Yeah, this makes it very intuitive. Makes it, yeah, the UI of extremely intuitive, it is.
- Yeah, so. And I was gonna say like, the reasons it's messed up is because the AI still isn't very good at picking up how to kind of do all of this. If you start from a template, so if I go, if you start from a template, a lot of these are really, really, and if I start it from here, actually you can also say, give me a report on website. Oops. If you start from a template, it's not as, it's not as messed up is what I was gonna say. Yeah, because it looks like now I'm redoing everything. It looks like I have to redo the assessment on everything, which usually is not the case. Usually most of our users would start from a template. That template, you know, would already be very accessible. You're already like 80% of the way there. It's not all over the place. You know, the accessories, you will not get like 50 errors like you saw before. You might get a few errors and then it's another five or 10 minutes of running the accessibility checker to essentially, you know, clean things up. And then you click on the export button and you're done, so.
- And Eugene, I'm assuming your customers, your users create all different types of infographics and like truly across the board, yeah?
- Yeah, so they're creating everything. I mean, it's not just infographics. So I mean they they're creating, I think like anything from posters to, so I think if I go back to templates, which is here, sorry. Yeah, you can see brochures, mind map posters, like even presentations. We're not literally a big presentation layer, but we do have presentations, flyers. Like, it kind of goes down. Like you can kind of create anything that you want. There's a whole bunch of them. We've sort of evolved from an infographic tool to like this all purpose AI design tool. And now we're sort of evolving into a, we're evolving into an AI design agent, if that makes sense. So that's kind of like the direction we're going into right now. So this is, for example, and if I go into the assessment, I'm quite sure it's a lot better now. It should be a lot better, yeah, so.
- [Eamon] There you go.
- [Eugene] So I can see this is, this is tagged correctly, H1, yeah, so. Actually this is not, should not be an H1, this should be an H2, but yeah, so. For the most part, a lot better than if you had just done it the other way around.
- [Eamon] When you use a template, yep.
- [Eugene] Yeah.
- Creators who have disabilities, how accessible is the creation of the tool? It's clear that the output, there's a lot of effort here to make sure that the output is accessible, but how is the actual experience? 'Cause some of that looks really hard to make accessible, all that visual stuff, getting that to.
- Yeah, so like, yeah, so for the most part it is keyboard navigable. But because it's drag and drop, a lot of it is, you know, it is very challenging to make accessible. So I would say, you know, it is accessible at a very basic, keyboard navigability is, for most of our screens, especially on this screen works well. Yeah, but if you're trying to, and for most of these widgets editing, most of these widgets work well as well because we do, we do have, you know, by default we do test for accessibility in the editing experience as well.
- And by the way, what is the testing processes for accessibility and what other procedures, like you've kind of put together as you did that shift some years ago to accessibility?
- Sure, so I think when we definitely, we implemented like automated testing for sure. Like, so there's automated testing for accessibility at the first level, and then all our QA now, you know. They have been trained to test with screen readers. So this is manual testing. So there's the automated part and then there's the manual part. We mainly use Nvidia, just Nvidia, because, I keep saying Nvidia, I dunno why.
- Because you're into AI and there's no bigger monster.
- Yeah. So they mainly test with, and they are trained with that. And so on the manual testing side, they also do that testing. And then on the requirement side, this happens most of the time. I have to say like, you know, there were some instances where it kind of felt, which it didn't happen, but which we found out last week. But for the most part, most of the requirements, you know, from when it comes from the design state, has accessible requirements attached to it. 'Cause you know, we do, for example, we do automated, you know, contrast checking, color checking, right? So if you're we're building a feature that touches color, we kind of have to also write the requirement on how that impacts the accessibility checker. And so for the most part, most of what, most of the requirement has a line, sorry, has some requirements that are specific to accessibility because it will break the accessibility checker. If we don't do it, it breaks the accessibility checker and we'll see all these errors in the accessibility checker. And so the process is sort of built in early on because it's part of the requirement and then the team kind of knows how to test it that way. And then we have done, so we have done also testing with end users as well. So we've done several rounds. We don't do it every, so we do, we release every week. We don't do it every week. I mean, we do it every couple of months. Every big, I would say when we get like a big release, we'll sort of recruit some users with disabilities and let them, you know, let them test it. You know, and we all, obviously, we always find issues and most of that testing is testing the user experience. It's not testing the PDF, because A PDF we can test fairly easily and fairly well. It's very well defined.
- Right.
- There's Pax checker, there's so much stuff we can do. It's mainly testing the UI and to see how accessible it is.
- [Joe] Sure.
- And yeah, so we do find a lot of, you know, we do find a lot of issues with, mainly with things that have been changed, like, you know, recently, like we'll have a change and we didn't understand the impact of it. Like, just 'cause, you know, whatever reason. It's, you know, so it's hard to, sometimes you just don't think of it and then you're like, oh yeah, when they click on this and this, like somehow, you know, the dropdown doesn't work, you know? Or it doesn't make sense, you know, if you're using.
- I think that's the key piece about the importance of still working with individuals and collaborating with individuals with disabilities. It really is because, yes, it's one thing to be conformant, but there's a usability level in there as well. That always has to be taken into consideration. There really is. So I'm happy to hear that that engagement's still ongoing. I do believe AI will help from a testing accuracy perspective, but I don't think it'll ever eliminate that engagement need. So, been 13 years now, I believe. overall Eugene, with Venngage and obviously a lot of growth. Anything you wanna share with our listeners as it relates to where you're going? Anything you can share as it relates to your roadmap?
- I mean, I think from a tool perspective, it's, you know, I seriously don't think we're that different from all the other tools out there. We are building, you know, more and more AI driven features and we are going towards, you know, building a design agent, which I'm sure everyone else is as well. And I think the interesting and exciting thing is that we, you know, AI has leveled the playing field. Like for example, Canva doesn't have a better image center than we do. Like we have the same, you know? They did buy a image, they did, they bought Leonardo, which I think it's okay, it's mid, it's not that great compared to Nano Banana and all these other image generators. So if in fact they're probably, you know, in a sense, you know, they have one hand tied behind their back now. 'Cause they have their own image model, but it's not as good as the state-of-the-art ones. So I'd say that AI has leveled the playing field, which is great. And which means we can now create features that we wouldn't be able to do before because we just didn't have the tech, right? Now we have access to the same tech as everyone else. So it's pretty exciting. I would say users, you know, at the end of the day end users, it's great for end users. 'Cause now you get a lot of smaller companies like ours who are able to create really good products, because we have access to all these models.
- Well, it's funny speaking about the giants versus the small businesses. I actually literally a few weeks ago was at the headquarters of Canva because I was in Australia to give a keynote. And they're the second biggest company in Australia. I didn't even know that they were based there.
- They're huge, yeah.
- And my hope is that AI is going to democratize things the same way that I hope that the web was going to democratize things and help uplift the small business, which in some ways it has, but in others, it went from this decentralized place to a centralized place where almost everybody essentially goes to these top five, six tech companies. So to that end, we will see what happens with AI. But I'm curious, are you using AI to build Venngage at this point? And if so, what percentage of the code is written by AI?
- So yes, we are using Venngage to build, to code as a coding. We are using coding agents. You know, how much of the new features are being, I would say every developer is using a coding agent right now. Every developer, not just in Venngage, probably everywhere else, I would assume. And how much of the, I mean, I don't know how much of the code is now written by. I would say definitely boilerplate, like simple code 80, 90% of that is probably all AI written. The ones that are more difficult to understand and more nuanced, that's probably still written by, by, you know, maybe 50% AI, 50% humans. And that change has only happened recently. And when I say recently, maybe six months ago. Even a year ago, it was still predominantly, I would say the agents were okay at like auto completion and just doing, you know, little checks here and there. But, you know, from last year to this year, the coding agents have improved so much that we're now generating far more code than, you know, like a lot more code, like 50% more probably. I don't really know the percentage we haven't measured, but I would guess at least 50% is now generated.
- Eugene, has your customer expectations changed in any way since AI?
- Yes. I think they all expect from click and then you get this like.
- Easy button, easy button.
- Yeah, yeah. And we can read their mind and you know, and then we can just give them what they. I think the expectation has changed a little bit, but I also think that a lot of people, so do see two different types of users. We do see the, I would say the AI native users, the ones who are students, the ones who kind of like prompting is very natural. So for me, like prompting is quite natural. I can write a paragraph long prompt. I have no issues with that. I've done it so much. But I would say that, so that's one group of users. And then you get this other group of users who are, let's call them non-native AI's, that prompting is very difficult. They struggle. They'll literally write like two words and then they expect the AI to understand those two or three words and create. So we're now working on trying to figure out like how can we ask for more color clarification? Can we ask them for like style preferences and all of that stuff. So to your question, yeah, expectation has gone up, but it's also created a bunch of complications in building a you know, in actually creating something the user wants.
- Yeah, that's fair.
- Interestingly, this might might be of interest to you. I just created over the weekend a whole bunch of skills to improve the under Claude code to improve the experience. And one of them is slash opt for optimizer. And what it does is it looks at whichever model I'm using, it pulls down, and well actually I pre pull it down, the model card as well as the prompt, any kind of document that typically the model company tells you on how to prompt it.
- Right.
- And what I have it do is I write the prompt and it rewrites the prompt and runs it by me. And so I think that the lot of alpha, as they would say, in prompt optimization, and I wish that more of these companies would just have a little button that lets you optimize the prompt.
- Okay.
- So if I were you and I'm trying to improve that, I would just have a button that says, all right, let's run this through a prompt optimizer and let it run.
- Yeah, so we, I dunno if we've released it. So we have, I def I definitely saw it in one of the product writers, basically, I think we call it prom enhancer or something like that. And it does something very similar to what you just mentioned. Basically it just improves the user prompt and then, and then, you know, most of the time it is better than what the user. And then you click on it and then you get a better result, because the prompt is better, right? So to your point Joe, I think it is a big problem. Is AI gonna actually democratize access to technology or is it actually gonna be worse, right? And I think, yeah, that's a tough one because I do see how you need so much money to actually build a model that in that, in that sense, the foundational models are definitely a consolidation. It's gonna be like, you know, Google, check, you know, it's gonna be like a handful of people, like five companies maybe that will own all the foundational models, which is worrying because, and you know, basically that drives all of the workflows of the world. I've got five companies owning all the workflows, right? So I think that is a big problem. On the other hand, it also allows like people like you and me to like create apps and create designs, create anything we want, where before, we wouldn't be able to do that. So I think it's a, you know, you get both right? You've got the very big companies who consol, who are definitely gonna consolidate a lot of that power and the economics of that. But it might also allow a lot of people to create new businesses, create new things, so.
- Yeah, that's, honestly, it's one of the reasons why we're doing this. Right between this and AMAC, like independent, if it's the five big boys or if it's a collective, multiple smaller organizations, we wanna make it accessible. We really do. Both from an LLM perspective and anything that generates perspective. So we talked about, you know, the early 2000's and the.com area is how this conversation started. It's when accessibility was a complete afterthought, as we all know. And we just can't make the same mistake again. We can't, we can't afford to. That's really the reason why we're doing this and meeting with awesome individuals like yourself Eugene. Before we wrap up, is there anything that we didn't touch upon that you wanna share with us? I know it's a very open-ended question, but I know we covered a lot.
- I think one interesting topic would just be like, will AI just make it unnecessary for us to worry about whether documents are accessible or not, or webpage is accessible or not? Because, you know, you can integrate, you know, vision, AI vision with which we, which they've done, you know? And it'll just read it the way humans read it. You won't have to, you know, who cares what's behind the code. Nobody, you know, like a webpage will just be read like a human would read their webpage. Doesn't matter if its got any area labeled, it doesn't really matter. It'll fill up the form the way you want it to fill out. 'Cause the AI agent will just be able to figure it out, right? So I think one of those big questions that I have is like, will AI just solve this problem for us? And, you know, I think it will probably solve like 80, 90% of the problem. I don't know, maybe that's the question for what you think, both of you? What do you think?
- I think if you answer that question, you get a lot of hate.
- One way or the other, we'll see.
- Yeah, we'll see what happens.
- I mean basically I, you know, I'm like, if they did that then they wouldn't eat my tool, right? Like, to be honest, they wouldn't, right? So, and not just my tool, like all the other accessibility requirements would be, you know, sort of not relevant anymore. But I do think for the, you know, but I'm not thinking like, just for myself, I'm thinking like for people who need to access this stuff, it would be great, right? You don't really, you don't have to rely on the creator to make it accessible. The screen reader itself or whatever, you know, accessibility tech you're using is just gonna be able to figure it out, right? And just read the webpage. You don't need to worry about whether the webpage is, you know, or the form is accessible. They're just gonna figure it out and just do it the way, you know, and just do the task for you, right?
- Yeah.
- So that's the big question I always, you know, I think about like, oh yeah, AI can just solve this problem for us eventually.
- I think it's very similar, the personal engagement's very similar to the business engagement that a lot of companies right now want to use AI. They know they need to use AI and it will help, but clearly determining exactly where they can mostly benefit from it and exactly when they'll get that ROI and exactly how trustworthy it really is, and what the risk level is, that's taking a little bit longer than expected. And I think that personal engagement will follow the same path. I really do, yeah.
- And don't want to dodge the question. So the way I answered earlier might have sounded like a dodge, but I'll put it this way. I've asked that question to three friends of mine who are blind. One of them is probably would be offended by the question as I was suggesting. The two others, and he knows AI a little bit like pretty well. And then two others know it really well and they both think that it will solve a lot of the problems.
- Right. And one of them is actually building something to solve the problem,
- Right.
- And I think that that's where it's gonna, it's gonna come from. But you know, when I say the problem, it's a huge host of problems, right? So you have the usability, which everybody finds that usability a problem to begin with. But then you have, all right, if you're using assistive technology, what kind of assistive technology are you using and what are your preferences? And hopefully the foundation model companies will have accessible platforms themselves, but if they don't, then somebody is gonna build something on top of it to make it accessible. So I think it's gonna improve a lot for the end user, but it's gonna be very challenging for industry in terms of jobs and things like that.
- Eugene as my favorite podcast, Smartless, will often say, we've taken up up enough of your time today. Thoroughly, thoroughly enjoyed the conversation. For our listeners that want to get in contact with you or learn more about yourself or Venngage, where would they find that?
- LinkedIn is probably the one place that I'm most active on.
- And you're Venngage.com? Yep.
- Yeah, and my email Just Eugene@Venngage.com Yeah, so.
- [Eamon] okay.
- Well thank you so much Eugene. It was really a pleasure.
- Oh, thanks over to you. Likewise.