When I got into crypto, I wanted to build a blockchain.
I had just read Bitcoin’s white paper and was mind-blown.
I marveled at the idea of a decentralized technology that didn’t allow centralized authorities to control trade and commerce.
A short while later, a buddy introduced me to trading.
At the peak of the last bull run, I made incredible amounts of money.
It was so easy that I dropped everything else and went all in.
Like every first-timer, I round-tripped most of the money.
Over the next couple of years, I continued making money but also stepped back into building.
In the last two years, I stopped trading entirely and made most of my crypto income from building.
But this year, things changed drastically, there were barely any jobs.
Thankfully, I had a cover: as a programmer, I wasn’t limited to building in crypto.
I started accepting Web2 development jobs once again and barely felt the difference.
A lot of people flock into crypto because of the promise of quick money.
I am especially concerned about the new entrants. Many of them complain that they aren’t making money and are frustrated.
Most of these are very young people between the ages of 15 and 25.
Some others, desperate for quick growth and the title of “KOL,” lie at every turn with fake posts and screenshots.
For context, I was a programmer for 9 years before I made serious money with my skill.
I was patient because I believed that my skills would eventually lead me to financial freedom if I kept being consistent and valuable.
The promise of quick wealth from crypto has ruined the mindset of so many young people.
The cars, houses, and other material acquisitions they see being flashed on the timeline have played into their psyche.
Here’s my advice for everyone in crypto who isn’t getting the results they expect yet:
•Get a skill and build real-world value.
•Do not rely entirely on crypto for income — but don’t leave either.
•Get a skill or job that will earn you money, and keep showing up.
Web3 will change your life, but it might not happen as magically and as quickly as you expect.
GM!
Introducing Claude Code Security, now in limited research preview.
It scans codebases for vulnerabilities and suggests targeted software patches for human review, allowing teams to find and fix issues that traditional tools often miss.
Learn more: https://t.co/n4SZ9EIklG
Anthropic had 16 AI agents build a C compiler from scratch. 100k lines, compiles the Linux kernel, $20k, 2 weeks.
To put that in perspective GCC took thousands of engineers over 37 years to build. (Granted from 1987 - however) One researcher and 16 AI agents just built a compiler that passes 99% of GCC's own torture test suite, compiles FFmpeg, Redis, PostgreSQL, QEMU and runs Doom.
They say they "(mostly) walked away." But that "mostly" is doing heavy lifting.
No human wrote code but the researcher constantly redesigned tests, built CI pipelines when agents broke each other's work, and created workarounds when all 16 agents got stuck on the same bug.
The human role didn't disappear. It shifted from writing code to engineering the environment that lets AI write code.
I don’t know how you could make the point AI is hitting a wall.
Happy birthday to Facebook.
One key lesson this photo teaches is: don’t build forever or wait for perfection.
Build an initial product that works, ship it, and improve it over time.
There will be divided opinions on this, but if you want to vibe code at speed, don’t use TypeScript.
LLMs are still not perfect enough to write code 100% bug-free or without syntax issues. From experience, most agent-generated TypeScript ends up with several type errors that eat up your time fixing.
So if you care about speed, or you’re an inexperienced coder, just use plain JavaScript. TypeScript compiles to JavaScript anyway, and there’s no performance penalty for using JS directly.
There are exceptions though. Some newer frameworks force TypeScript. Expo, for example, doesn’t support plain JavaScript in recent versions.
As you vibe code your apps, you also need to be aware of performance issues caused by bad code.
You might have started vibe coding that app just for fun, but if it goes viral or turns into a serious product, performance will matter a lot.
Speed is everything in UX. If your app isn’t fast, it can easily kill that beautiful app you vibe coded once it starts to take off.
Claude is great at coding. Like really, really good, compared to using a regular LLM like Gemini.
There's only one issue with it. Because it never sees the full code of the app but uses grep search for relevant code snippets, it's myopic.
If grep returns a fragment of code similar to the bug description, it often doesn't look further and fixes an irrelevant part of the app or answers a question based on these fragments found by grep.
So, as the codebase grows, it becomes important for the user to know the codebase. Otherwise, Claude will reinvent the bicycle over and over again, creating duplicate implementations for the same functionalities in different places in the app.
This issue is probably fixable with additional finetuning, but right now this is how it works.
One single prompt I constantly use to prevent performance bottlenecks in my vibe coded apps:
"Check my codebase for memory leaks and recommend fixes"
After it finds them, ask the agent to fix.
Do it repeatedly a couple of times and your code would definitely be ridded of any code issues that can cause performance issues.
As you vibe code, be aware of what’s called a race condition.
It happens when multiple actions run at the same time and the result depends on which one finishes first.
This can cause users to get more value than they should, and potentially cost you real money.
It’s one of the most critical bugs in any app, and the consequences can be disastrous.
AI generated code is not immune to bugs. Just like humans, it can make mistakes.
The security gap in vibe coded apps isn’t inherent to vibe coding itself, it’s a problem of wrong practice.
Software passes through several stages before reaching production, and in every team, different people are responsible for each stage.
The good part is that AI can be your team.
Instead of vibe-coding an app and sending it straight to production, create sub-agents that perform the same tasks a human development team would handle before launch.
That’s how I build industry grade apps that outperform without compromising quality.
Watch out for anyone who feels vibe coding is bullshit and is constantly saying AI coding is rubbish.
Those are people who feel they are better than everyone else.
That's a huge psychopathic trait.
One thing no one has been able to fight is change!
Every naysayer of AI will regret it in years to come!
This is just an attention seeking post.
I have coded for 16 years and nothing comes close to AI assisted coding.
You can't travel with your legs to a different continent because you are in good health, you must fly or use an assisted means of transportation.
I have multiple apps in production built with claude code and I can tell anyone who cares to listen, nothing beats Opus 4.5
Every naysayer of AI will regret it in years to come!
This is just an attention seeking post.
I have coded for 16 years and nothing comes close to AI assisted coding.
You can't travel with your legs to a different continent because you are in good health, you must fly or use an assisted means of transportation.
I have multiple apps in production built with claude code and I can tell anyone who cares to listen, nothing beats Opus 4.5