Mortgage Bankers Are Vibe Coding Now
Vibe coding went from Karpathy tweet to Stanford syllabus in thirteen months. When mortgage bankers are learning it, the tipping point is ancient history.
The Mortgage Bankers Association held a vibe coding workshop on March 12th. I had to read that twice. Not a fintech startup. Not a dev bootcamp. The Mortgage Bankers Association. Teaching people who process home loans how to talk to AI and build software.
Andrej Karpathy tweeted the phrase “vibe coding” in February 2025. Thirteen months later it’s a Coursera specialization. Stanford Continuing Studies is teaching it. MIT added it to their Missing Semester course. Thirteen months from shitpost to syllabus. For comparison, object-oriented programming took a decade to reach university curricula. Functional programming still hasn’t made it to most of them.
Something is moving faster than I think most people have internalized.
The Week Everything Showed Up at Once
This is the part that got to me. It wasn’t one signal. It was all of them landing in the same week.
The New York Times Magazine published “Coding After Coders,” a 10,000-word piece where Clive Thompson talked to 70+ engineers at Google, Amazon, Microsoft, and Apple. His finding: engineers now describe requirements and approve AI-generated plans more than they write code. The NYT doesn’t commission 10,000-word magazine features about fads. That’s the paper of record saying the shift is real and permanent.
Same week: Coursera’s specialization went live. Covers GitHub Copilot, Cursor, Claude Code, MCP, ChatGPT. Udemy has 7+ dedicated courses. Andrew Ng launched 4 free ones. Google announced AI literacy training for 6 million US educators. OpenAI and Google are both racing to launch competing certification programs.
And the mortgage bankers. The mortgage bankers showed up to learn.
When the most conservative industry in American finance starts teaching its people to build software with AI, you’re not witnessing the tipping point. You’re witnessing what comes after.
The Part That Made Me Laugh
There’s a tool called Einstein that connects to Canvas LMS and takes entire courses autonomously. Passes assessments. Submits assignments. Earns certifications. No human involved.
So Coursera launched a vibe coding specialization, and the thing the specialization teaches you to use can already complete the specialization on its own. The credential is dead before the first cohort finishes.
I keep turning this over in my head. The courses teach you to prompt AI and review its output. The AI is good enough to prompt itself and review its own output. What exactly is the course certifying? That you can do something a machine does better?
Universities are going to have a rough few years figuring out what to do about this. Proctoring won’t fix it. The problem isn’t cheating. The problem is that the skill being credentialed is the same skill being automated.
What’s Actually Worth Learning
This is where I’ve landed after thinking about it for a while. There’s a real distinction forming between vibing and engineering, and the line matters more than people realize.
Vibe coding gets you 80% of the way to something useful. Describe what you want, iterate on what comes back, ship it. Craig Mod, a writer with no engineering background, built a full multi-currency accounting app in 5 days with Claude. Handles US and Japanese tax systems, reconciles international wire transfers. That’s not a toy. A non-engineer built serious production software by vibing.
For a mortgage banker building an internal calculator, 80% is more than enough. For a founder shipping an MVP, it’s a gift. For someone who needs the other 20%, who needs to understand why the system fails at scale, how the security model breaks, where the architecture won’t hold, vibing isn’t enough. That last 20% still requires understanding things no 13-month-old course can teach.
Simon Willison keeps making this point and I think he’s right: the goal should be demonstrably better software, not just faster software. Knowing when to vibe and when to engineer is the skill that actually matters. The difference between “I built a thing” and “I built a thing that works at 3am when nobody’s watching.”
Where This Goes
Vibe coding entering mainstream education means writing code by hand is going the way of writing assembly. It still happens. Specialists still do it. It’s not the default anymore.
The mortgage bankers will build tools their IT departments would have deprioritized forever. The Stanford students will enter a workforce where conversational programming is table stakes. The Coursera grads will have a credential that proves they can do what everyone can already do.
The courses will mint a million people who can get AI to generate code. The market will sort out which ones know what to do when the code is wrong.