As Coinbase rapidly adopts AI, we have a responsibility to bake accessibility into our development lifecycle from the start. This talk addresses clear accessibility gaps in AI, from biased algorithms to inaccessible outputs, and presents our proactive strategies to mitigate them. We will cover our work building with accessibility in mind via AI. This session is a call to action to ensure that building with AI doesn’t mean shipping inaccessible experiences.

Proactive Inclusion: Embedding Accessibility into the AI Revolution at Coinbase, outlines how Coinbase is moving beyond reactive accessibility by integrating AI into the heart of their development lifecycle. It focuses on shifting from manual auditing to an AI-partnered model that ensures inclusive experiences are built by design rather than as an afterthought.

Key Takeaways

• Educate & Streamline: Train AI agents on specialized accessibility knowledge bases to provide 24/7 expert guidance and automate redundant tasks, such as CEFR language proficiency reviews and audit documentation.

• Fix & Create: Empower accessibility practitioners to use AI agents (like Cursor and Claude) to bypass engineering bottlenecks, allowing them to directly write accessible code and build internal tools.

• Automate & Review: Utilize AI-driven workflows, such as the RAID (Retail AI Developer) GitHub Action, to automatically evaluate pull requests for regressions and review designs for accessibility before they reach production.