In partnership with
AI Agents Are Reading Your Docs. Are You Ready?
Last month, 48% of visitors to documentation sites across Mintlify were AI agents, not humans.
Claude Code, Cursor, and other coding agents are becoming the actual customers reading your docs. And they read everything.
This changes what good documentation means. Humans skim and forgive gaps. Agents methodically check every endpoint, read every guide, and compare you against alternatives with zero fatigue.
Your docs aren't just helping users anymore. They're your product's first interview with the machines deciding whether to recommend you.
That means: clear schema markup so agents can parse your content, real benchmarks instead of marketing fluff, open endpoints agents can actually test, and honest comparisons that emphasize strengths without hype.
Mintlify powers documentation for over 20,000 companies, reaching 100M+ people every year. We just raised a $45M Series B led by @a16z and @SalesforceVC to build the knowledge layer for the agent era.
A few days ago, engineer Kanav posted a meme that spread across X very quickly.
The meme showed a developer giving an AI assistant full access to a codebase. Then the developer asks a simple question.
"Who is JSON?"
The joke is funny because JSON is not a person.
It stands for JavaScript Object Notation. It is one of the most basic formats used to exchange data between applications. Almost every developer has worked with it.
The meme was making fun of a growing trend called vibecoding.
Then Elon Musk saw the post.
He quoted it with just two words.
"Full access"
That reply exploded.
The original post crossed tens of thousands of likes. Soon people started sharing their own stories about blindly trusting AI. Someone even launched a $JSON memecoin on Solana. For a short time, the token surged more than 500 percent.
A simple joke had turned into one of the biggest conversations in tech this week.
But behind the laughs lies a serious question.
Are developers becoming too dependent on AI?
What is Vibecoding
If you spend time on X or YouTube, you have probably heard this term.
Vibecoding means describing an idea in plain English and letting AI write most of the code.
You do not think too much about syntax.
You do not spend hours reading documentation.
You simply tell the AI what you want.
"Build me a login page."
"Create a chatbot."
"Make an app that tracks my expenses."
The AI writes the code.
You copy it.
You run it.
Sometimes it works.
Sometimes it does not.
The term became popular after Andrej Karpathy talked about this new style of programming in 2025.
The idea spread quickly.
Because honestly, it feels magical.
A beginner can build an app in a weekend.
A designer can launch a website without hiring a developer.
A startup founder can create a prototype before even writing a business plan.
AI removed many barriers.
That is a huge win.
But it also created a new problem.
When you stop understanding your own code
The "Who is JSON?" meme is funny because it exaggerates a real fear.
What happens when developers use AI without understanding the basics?
Imagine giving an AI assistant access to your codebase.
It changes files.
Updates databases.
Creates APIs.
Deploys new versions.
Everything looks fine.
Then something breaks.
Users cannot log in.
Payments fail.
The server crashes.
Now you have to fix it.
But what if you do not understand the code AI wrote?
What if you do not know how APIs work?
What if you do not know what JSON is?
Suddenly the magic becomes a problem.
You are responsible for a system you do not fully understand.
That is scary.
AI is great at speed
Humans are great at judgment
This is something many experienced developers are learning.
AI can generate code very fast.
It can write functions in seconds.
It can explain concepts.
It can find bugs.
It can even suggest better solutions.
But AI does not own the consequences.
You do.
If an AI assistant deletes a database, the company loses money.
If an AI creates a security flaw, customers are affected.
If an AI writes poor code, your team has to maintain it.
This is why professional developers are not giving AI complete freedom.
Most of them use AI as a partner.
Not as a replacement.
They ask questions.
They review the code.
They test everything.
And most importantly, they keep learning the fundamentals.
Because understanding matters.
The calculator lesson
Think about calculators.
When calculators became common, some people worried that students would forget math.
And to some extent, that happened.
Many people can multiply numbers.
But fewer people understand why formulas work.
AI could create the same situation for programming.
Developers may become very good at prompting.
But weaker at reasoning.
Very good at generating code.
But weaker at debugging.
Very good at moving fast.
But weaker at building reliable systems.
The danger is not that AI becomes smarter.
The danger is that humans stop learning.
Yet vibecoding is not bad
It is easy to laugh at vibecoding.
But the truth is more complicated.
This new way of building software is helping millions of people.
Students can learn faster.
Entrepreneurs can test ideas cheaply.
Small businesses can build tools without huge engineering teams.
Creativity is exploding.
People who never wrote a line of code are now building games, websites and apps.
That is amazing.
We should celebrate it.
The problem starts only when people believe they no longer need to understand anything.
Because eventually, reality catches up.
Servers fail.
Users complain.
Security issues appear.
And at that moment, prompts alone are not enough.
You need knowledge.
You need experience.
You need judgment.
The future belongs to builders who know both
The best developers of the future may not be the people who reject AI.
And they may not be the people who blindly trust it either.
The winners will probably sit somewhere in the middle.
They will use AI to move faster.
But they will also understand the fundamentals.
They will know when AI is right.
And more importantly, they will know when AI is wrong.
That combination is powerful.
Because AI can generate thousands of lines of code.
But only humans can decide if that code should exist.
One meme. One lesson.
The internet will probably forget the "Who is JSON?" meme in a few days.
The memecoin will disappear.
People will move to the next joke.
But the lesson will stay.
AI is becoming incredibly powerful.
Giving it full access can feel exciting.
Sometimes it even feels like magic.
But magic without understanding is risky.
So use AI.
Experiment with vibecoding.
Build weird things.
Ship products.
Move fast.
Just do not forget to learn the basics.
Because one day, when something breaks, you do not want to be the person asking
"Who is JSON?"
—Sushila