Podcast Summary
Episode Overview
Podcast: How I AI
Host: Claire Vo
Guest: Joe McCormick, Principal Software Engineer at Babylist
Episode: How this visually impaired engineer uses Claude Code to make his life more accessible
Date: February 16, 2026
This episode of "How I AI" delves into practical, personal software creation, focusing on accessibility. Joe McCormick, who is visually impaired due to Lieber's hereditary optic neuropathy, shares how he leverages AI—especially Claude Code—to create Chrome extensions that make both his work and life more accessible. Joe demonstrates his workflow, showcases custom-built tools, and discusses how the ability to rapidly prototype personal software is transforming accessibility—not just for himself, but for anyone willing to build.
Key Discussion Points and Insights
Joe’s Journey to Accessible Software Engineering
- Background on Losing Vision
- Joe describes his diagnosis before college and initial adaptation using old assistive tech (magnifiers, glasses).
- Shifted from robotics/mechanical interests to computer science at Harvard due to increased accessibility and creative parity (02:58).
"I got the same feeling of creativity and being able to come up with the idea and make it happen. But now I was on maybe on a full equal plane to my competitors [...] as AI took off became even more equivalent." — Joe McCormick [03:31]
- Impact of Modern AI Tools
- AI models (ChatGPT, Gemini, Claude) have dramatically narrowed the accessibility gap for visually impaired engineers.
- Sharing screens or asking for help is now much simpler thanks to AI's multimodal capabilities.
The Power and Value of Personal Software
- Lack of Custom Accessibility Tools
- There's little economic incentive for companies to create niche accessibility tools, but AI lets individuals build their own custom solutions (04:51).
"What I love about what we're able to do now with AI is not only are more interesting accessibility tools and platforms able to be built, but people can build these solutions for themselves." — Claire Vo [05:33]
- There's little economic incentive for companies to create niche accessibility tools, but AI lets individuals build their own custom solutions (04:51).
- Rapid Iteration and Personalization
- Joe highlights how the ROI for personal tools has collapsed: what used to take days now takes minutes (13:46, 32:56).
Showcase: Joe’s Custom Chrome Extensions
1. Image Description for Slack
(Demonstrated at 06:22 – 09:59)
- Problem: Parsing images in Slack is tedious using zoom/magnifiers.
- Solution: Hotkey-triggered Chrome extension (Ctrl + Shift + D) that uses AI to describe an image, and answer follow-up queries (e.g., "what age child is this for?"). No need to rely on teammates for image descriptions.
"I use this tool... to be able to get the gist of an image without needing to ask somebody to explain it to me." — Joe McCormick [02:22, repeated at 06:22]
- Efficiency: Built in ~25 minutes using Claude Code. Version for Figma in development (10:05).
2. Universal Spell Checker
(10:50 – 13:46)
- Problem: Typing with errors is common; existing tools like Grammarly are not always screen reader accessible.
- Solution: Custom spellchecker for any input box, triggered with Ctrl + Shift + S, integrates with screen reader (announces “spellcheck complete”) and only corrects typos while leaving content unchanged.
"They're not all screen reader accessible. They're multiple clicks away sometimes. So I built one out that works in any input field on the web." — Joe McCormick [10:52]
- User Experience: Highly efficient—shortcuts over mouse clicks; designed around real accessibility/UI need.
3. Link Summarizer for Slack
(14:37 – 41:00)
- Problem: Slack messages often contain links saved “for later” but rarely revisited; wastes time.
- Solution: Chrome extension that, with a keyboard shortcut (Ctrl + Shift + 1), finds links in a Slack message, fetches the article, uses OpenAI to summarize 3–5 main takeaways, and presents them in an accessible modal (15:58, 16:55, development at 16:13–41:00).
- Live Build Demo: Joe demonstrates building the prototype from scratch, including:
- Dictating a PRD via VS Code’s voice input
- Using Claude Code (including Claude Skills for Chrome extensions)
- Setting up config file symlinks for API keys (23:09)
- Iteratively refining the output and fixing display bugs with screenshots and custom slash commands
- Accessibility: Modal is fully screen reader accessible using ARIA roles, auto-focus, and avoiding reading content behind the modal (40:09).
"If you tell some of the foundational models like 'please make this accessible,' ... they do actually a great job of this." — Joe McCormick [40:21]
Notable Workflow Tips:
- Run Slack in Chrome: Allows extensions to modify Slack UX (12:07).
- Keyboard Shortcuts: Critical for speed and accessibility; latency is the “killer feature” (12:07, 26:37).
- Efficient Prototyping: Using skills and frameworks speeds up the process dramatically (29:07, 32:56).
Notable Quotes & Memorable Moments
- On Equitable Creativity:
"The gap between... a sighted person and a visually impaired person is closing day by day [...] it's even been extra impactful." — Joe McCormick [03:31]
- On ROI of Custom Tools:
"The payback period has just become insane for a lot of this tooling." — Joe McCormick [13:46]
- On Accessibility for All:
"Things that are good screen readers probably are going to be good tools for everybody across the board." — Joe McCormick [15:58]
- On Building Efficiently:
"You get compounding effects by staying in the same technical space when you're trying to learn these Vicod tools. Because you're not trying to learn on two fronts." — Claire Vo [31:30]
- On Using Multimodal AI for Parenting:
"A big shout out to the Gemini app and its live share features. I can now read any book. It sounds like me reading it, but me and my 3 year old Cole will sit on the couch, he'll bring a book over and I'll be like, hey, I can read this one. Or no, Gemini can read this one for us." — Joe McCormick [45:08] "Now that 'sorry I can’t' becomes like 'sorry, I can' with the assistance of so many different tools now." — Joe McCormick [45:45]
- On Managing AI When Stuck:
"My typical mode is basically clear the context and start fresh as much as possible. … Something clearly got poisoned in this context. And when you start from scratch, I feel like everything just feels smoother." — Joe McCormick [47:18]
Key Timestamps
- 00:00 – 02:58: Joe’s backstory & journey into accessible engineering
- 06:22 – 10:05: Demo of image description extension for Slack & Figma
- 10:50 – 13:46: Spellchecker extension & efficiency workflow discussion
- 14:37 – 41:00: Step-by-step live build of Slack link summarizer extension; accessibility features and rapid prototyping techniques highlighted
- 42:59 – 46:33: Discussion on MCPS (multi-channel personal software), file translation, and impact on accessibility
- 45:08 – 46:33: Personal note on using AI to read to his children—Gemini app use case
- 47:18 – 47:57: Joe’s strategy for dealing with stuck AI sessions: reset context
Additional Insights & Takeaways
Accessibility and Mainstream Software
- Many large enterprise SaaS apps are becoming more accessible (Google Docs is highlighted for attention to detail), but complex apps like Notion still have challenges.
- Multimodal tools allow visually impaired users to access content and features in new ways—translating images to text and documents to markdown for easier screen reader navigation.
Meta-workflows and Tooling
- Joe automates even meta-workflows (e.g., screenshot importing across OS boundaries, config file symlinks) to minimize “microfriction.”
“I added a slash command for Paste Image... again this is similarly I would copy a file and I'd like save it in Windows, then move it to Linux... and I was like, there's gotta be a better way.” [37:35]
- Domino effect: Each new extension (the third, fourth, fifth) takes even less time to build, due to reusable skills and frameworks.
Resource & Contact Info
- Joe McCormick: On LinkedIn, happy to connect re: accessibility or Chrome extensions (48:17)
- Babylist: Hiring engineers, particularly those comfortable building with AI
Episode Summary
Joe McCormick demonstrates how AI—specifically, tools like Claude Code—enables visually impaired engineers (and everyone) to rapidly prototype highly personal, efficient accessibility tools that transform daily work and life. By walking through his real-world workflow, Joe illustrates not only how to make personal software with modern AI assistants, but how building such tools is now practical and offers a staggering ROI. The episode is a masterclass in blending empathy, technical skill, and the very latest AI to create a more accessible world, one keyboard shortcut at a time.