As I wrote yesterday in my story on the newly-released PBS Kidsâ series Lyla In The Loop, the networkâs senior vice president and general manager, Sara DeWitt, makes fairly frequent appearances in this columnâs coverage. The Palo Alto-based digital accessibility company Evinced is yet another recurring character, as Iâve interviewed founder and chief executive officer Navin Thadani several times here over the years about various topics pertaining to his companyâs work.
Earlier this week, I sat down not with Thadani but his colleague in Evincedâs chief marketing officer, Wade Lagrone, about the companyâs announcement made today that its customer base tripled in 2023 on the backs of software tools such as Unit Tester and Design Assistant. Both tools were developed to âautomatically [detect] the most difficult accessibility problems long before they are released to consumers.â Moreover, Evinced said in its press release the aforementioned tools âhave strengthened the companyâs position as a leader in web and mobile accessibility technologyâ which enables them to partner with big-name organizations (in the United States) like Capital One and Verizon.
In a nutshell, the crux of my conversation with Lagrone was on user testing. Lagrone characterized the accessibility industry, which includes Lagrone and his comrades at Evinced, at the precipice of a âsea changeâ involving how accessibility testing is conducted. Lagrone explained how, over the last few decades, engineers would run tests of accessibility on a website or something, write up a report, and forward it to the team so they could address the issue(s). The glaring problem with this model, as AudioEyeâs Mike Paciello told me last week, is it doesnât at all scale well. Itâs too slow. These audits, as theyâre called in the business, are outmoded because improvement doesnât come nearly fast enough.
âThat [existing] model canât keep up with a modern, large enterprise. If you are a mom-and-pop website and you make changes to your website once a month or once every couple of months, maybe that would works,â Lagrone said of the status quoâs liabilities in accessibility. âEven then it would be hard. But for a modern software development organization, it was never going to workâand it doesnât work. What youâve seen in the industry in the past is [people] trying to figure out how to move that testing process earlier in the lifecycle of the development process.â
Lagrone went on to emphasize that the story isnât so much that Evinced is growing exponentially for a three-year-old startup. Rather, the real story is Evinced is seeing growth is because the idea for modernizing in software development processes is rapidly gaining steam. Put another way, Lagrone and team are effectively trying to get the wider world to notice that prioritizing accessibility as a constant presence during development is exponentially betterâand pointedly, much more efficientâthan treating it as this extraneous, ancillary thing glommed onto the product right at the end when something is scheduled to ship.
âWeâve been laser-focused as an organization from the get-go on what we call critical accessibility problems,â Lagrone said of Evincedâs modus operandi. âIf you run an open source site scanner on your website, most of the issues that itâs going to report are relatively small accessibility issues [such as] missing alt-text and color contrast problems. These are problems weâre solving, but from our perspective, theyâre not critical.â
Critical to Evinced is stuff like the accessibleness of UI elements such as dropdown menus, that “no one else could solve automatically,â according to Lagrone. To this end, he said Evinced has leaned heavily, over years, on artificial intelligence and machine learning to solve this otherwise intractable technical problems. âWhat we do is sit around and think about all the time is how to get to perfection,â Lagrone said. âIf we can do it, if we can help people get there, weâre going to help them get there in the easiest way possible. Our belief has always been that if we can get to perfect in the easiest way possible, then itâll happen.â
When asked to expound on the companyâs views of AIâs rise in prominence in recent times, Lagrone reiterated Evinced has used the technology for a while in its products. He also noted most people think of AI in the ChatGPT-style form whereby somebody types in a query on command into a system and out pops information, kinda like a highly technical, highfalutin fancy-pants oracle. That use case doesnât apply itself well to solving for accessibility, Lagrone said, at least not the issues Evinced exists to tackle. Evinced is but one example of a company using AI for genuine good vis-a-vis accessibility, with Lagrone saying the companyâs ethos has long been to âconstrainâ artificial intelligence such that âthe answers it gives are correct and correct all the time.â
âAs generative AI has come along and the models have improved, we started looking at them as a way to help with accessibility,â Lagrone said. âOur realization has been they need a fair bit of corrective measures before they can be reliably used.â
Feedback has been positive, Lagrone said. He noted confidently the results âspeak for themselves,â adding the growth Evinced has seen suggests tools like Unit Tester and Design Assistant have proven instrumental in propelling the company towards its current position.
As to the future, Lagrone told me the ultimate goal is to someday get to a place where bugs donât happen and accessibility is persistent throughout development. To be forever bug-free sounds fanciful, but the salient point is Evinced is moving towards a day where accessibility is part and parcel of the development cycleâas it well should be, deservedly so.
âThatâs what we are aiming for. Thatâs why we get out of bed in the morning, which is to solve this because we think it is solvable,â Lagrone said. âWeâre making a tremendous amount of progress on that and weâre excited about it. Itâs gratifying to see that light at the end of the tunnel.â