Mckay Wrigley's AI Coding Education

Table of content

Mckay Wrigley built Chatbot UI — with 33k+ GitHub stars, it’s one of the most popular open-source interfaces for interacting with AI models. Now he runs Takeoff, an education platform teaching 25,000+ people how to code with AI tools like Cursor and Claude Code.

His trajectory captures something interesting about where AI development is headed: from building tools, to teaching people how to use tools, to teaching people how to build with AI assistance.

Chatbot UI: democratizing AI access

Chatbot UI started as a simple premise — give people a clean, self-hostable interface to talk to any AI model. No vendor lock-in. Your data stays yours.

The project exploded because it solved a real problem: most people couldn’t (or didn’t want to) pay for every AI subscription. With Chatbot UI, bring your own API key and you’re done.

33k stars later, it’s been forked nearly 10k times. That’s not just popularity — it’s people taking the code and building their own versions. The open-source approach created an ecosystem.

The pivot to education

Building tools taught Wrigley something: most people struggle not with the tools themselves, but with knowing how to use them effectively. The gap isn’t access — it’s knowledge.

Takeoff emerged from this insight. The platform focuses on practical AI coding skills:

Cursor mastery — moving beyond basic autocomplete to full agentic workflows
Claude Code — using AI for multi-file changes and complex refactors
Project-based learning — building real things, not toy examples

The courses get updated constantly because AI tools change constantly. What worked six months ago might be obsolete.

Teaching philosophy

Wrigley’s approach is explicitly beginner-friendly. His stated mission: “onboarding the world to AI.” Not just developers — everyone.

This matters because the AI coding tools are accessible enough that non-programmers can build useful things. Takeoff leans into this. You don’t need a CS degree to start.

The flip side: even experienced developers benefit. AI coding is a different skill than traditional coding. The mental models don’t transfer cleanly. Senior devs struggle with knowing when to trust AI output, when to intervene, how to structure prompts.

Open source contributions

Beyond Chatbot UI, Wrigley maintains several useful projects:

claude-code-voice — voice control for Claude Code
simple-multi-agent — minimal multi-agent implementations
takeoff-linear-mcp-server — MCP server for Linear integration

The pattern: build something useful, open-source it, teach people how to use it. Repeat.

What I take from this

Wrigley represents a specific archetype: the builder who becomes the teacher. He didn’t start with education credentials. He started by building things people wanted, figured out what they struggled with, then systematized that knowledge.

The Takeoff model — continuously updated courses on fast-moving tools — might be what AI education has to look like. Static content doesn’t work when the tools change every month.

And Chatbot UI showed something important: even in the age of AI, there’s value in giving people control. Self-hosted, bring-your-own-key, your data stays local. That philosophy scales beyond chat interfaces.