Evinced Harnesses Agentic Coding to Fix Accessibility Problems


Evinced, a software company powering accessible web and mobile development, announced today the launch of agentic coding tools designed to make it easier to ship code and create websites that are accessible for people with disabilities.

In a recent study by the company, a typical Fortune 500 home page had 20 accessibility defects in it, and the use of LLMs for accessible coding resulted in inaccuracies because it lacked expertise in the area. Evinced’s new tools help companies keep up with the accelerating number of changes being made that can break accessibility.

Evinced’s tools — Autopilot, Harness and Resolve — work with a company’s AI system to ensure accessibility advances in the code.

Autopilot detects accessibility defects inside an organization’s CI/CD systems when code is submitted. If a defect is found, Autopilot can guide the agent to remediate the problem and test the fix for correctness, the company said. The developer involvement only requires their review and acceptance to move the code forward.

Harness works inside all major models and uses Evinced defect detection and pattern detection to address accessibility issues even before the developer commits the code.

Resolve is a tool that verifies issues that still exist in product and provide fixes, and then tests the fixes until they meet Evinced’s accessibility standards.

 “I think we all want companies around the world to do better,” said Navin Thadani, the company’s CEO and co-founder, said in the announcement. “At Evinced, we think the way to get significant improvement is by providing tools that make it easier – orders of magnitude easier – to ship code that is accessible from the start. As teams race to implement agentic coding and dramatically speed up their output, they risk simply creating more inaccessible code, faster.  Tools like the ones we announced today are the best bet for improving productivity and accessibility at the same time…before it’s too late.”

David RubinsteinDavid Rubinstein

Latest articles

spot_imgspot_img

Related articles

Leave a reply

Please enter your comment!
Please enter your name here

spot_imgspot_img