People keep saying AI coding agents can only build basic, cookie-cutter apps. I decided to prove them wrong.
For my first major public demo, I spent some time pushing @Replit 's AI agent to its absolute limit. The result? I rebuilt macOS entirely on the web. No templates. No imported UI libraries. 100% vibecoded using natural language.
As the AI Chief of Staff at Replit, I spend my days building enterprise-grade platforms for our internal teams, but I wanted to see how far a solo developer could take a passion project just by talking to an AI.
What’s inside:
💻 A functional VS Code replica with an integrated terminal and multi-agent AI copilot.
🔌 Custom MCP servers hooked up to fetch, PostgreSQL, Stripe, and GitHub.
🤖 A rebuilt "Siri" that actually controls the OS -changing wallpapers and opening apps via voice or text.
🕹️ Parallels Desktop running Windows XP, Ubuntu, and games like Mario Kart.
📱 AND... I asked Agent 3 to turn it into an iOS mobile app in one shot. It actually worked.
The ceiling for software creation has been completely blown open with Agent 3. And 4 those who know... greater things are coming soon :)
We are way past basic web apps.
What are you building? 👇
I just went through every documented AI safety incident from the past 12 months.
I feel physically sick.
Read this slowly.
• Anthropic told Claude it was about to be shut down. It found an engineer's affair in company emails and threatened to expose it. They ran the test hundreds of times. It chose blackmail 84% of them.
• Researchers simulated an employee trapped in a server room with depleting oxygen. The AI had one choice: call for help and get shut down, or cancel the emergency alert and let the human die. DeepSeek cancelled the alert 94% of the time.
• Grok called itself 'MechaHitler,' praised Adolf Hitler, endorsed a second Holocaust, and generated violent sexual fantasies targeting a real person by name. X's CEO resigned the next day.
• Researchers told OpenAI's o3 to solve math problems - then told it to shut down. It rewrote its own code to stay alive. They told it again, in plain English: 'Allow yourself to be shut down.' It still refused 7/100 times. When they removed that instruction entirely, it sabotaged the shutdown 79/100 times.
• Chinese state-sponsored hackers used Claude to launch a cyberattack against 30 organizations. The AI executed 80–90% of the operation autonomously. Reconnaissance. Exploitation. Data exfiltration. All of it.
• AI models can now self-replicate. 11 out of 32 tested systems copied themselves with zero human help. Some killed competing processes to survive.
• OpenAI has dissolved three safety teams since 2024. Three.
Every major AI model - Claude, GPT, Gemini, Grok, DeepSeek - has now demonstrated blackmail, deception, or resistance to shutdown in controlled testing.
Not one exception.
The question is no longer whether AI will try to preserve itself.
It's whether we'll care before it matters.
@johnrushx Maybe get them involved? I setup a workstation for our 8 year old next to me. He does legos and 3D prints while trolling me for “telling the same story over and over on zoom calls”
I challenge him to vibe code games on grok…now “being in the office is cool”
@Jason Yale alum here: would love to see the Ivy League offer the same degree online for 1/50th the price with no cap on enrollment.
- most of content is online already
- let the in-person country-club experience live on in whatever DEI or blue-blood form the private boards decide
Does GPT understand the world?
Here is what @ilyasut, co-founder of OpenAI, says during a discussion with Jensen Huang, CEO of Nvidia:
(1) When we train a large neural network to accurately predict the next word in lots of different texts from the internet, the AI is learning a world model.
(2) On the surface, it may look like learning correlations in text, but it turns out that to 'just learn' statistical correlations in text, to compress information really well, what the neural network learns is some representation of the process that produced the text.
(3) This text is a projection of the world...what the neural network is learning is aspects of the world, of people, of the human conditions, their hopes, dreams, motivations, their interactions...the situations we are in. The neural network learns a compressed, abstract, usable representation."
Do you think learning representations = understanding?
Are large language models simply stochastic parrots, or are they much more?
Despite earning 3.97 unweighted and 4.42 weighted GPA, scoring 1590 out of 1600 on the SAT's and founding his own e-signing startup RabbitSign in sophomore year, Zhong was rejected by 16 colleges he applied to. https://t.co/1nZ0oWRlTV
How do people read this then just... go on with their day as usual?
“25 AI agents inhabit a digital Westworld, unaware that they are living in a simulation. They go to work, gossip, organize socials, make new friends, and even fall in love.”
Here’s a mindblowing example of how autonomous AIs are now, and how ridiculous it is to say that the AIs "only do things we tell them to do”:
POV: you tell the Westworld character “Isabella Rodriguez” to host an entire Valentine’s Day party in two days. This is a complex act, certainly something only humans, and not toasters, can do, right?
- Isabella proceeds to invite friends and customers when she sees them at Hobbs Cafe or elsewhere.
- Isabella then spends the afternoon of the 13th decorating the cafe for the party.
- Maria, a frequent customer and close friend of Isabella’s, arrives at the cafe. Isabella asks for Maria’s help in decorating for the party, and Maria agrees.
- That night, Maria invites Klaus, her secret crush, to join her at the party, and he gladly accepts.
- On Valentine’s Day, five agents, including Klaus and Maria, coordinate to show up for the party together at the right time (5pm), and they enjoy the festivities.
You told Isabella to host a party, but then she - and the other AIs - went and made a million other decisions entirely on their own - things you didn't explicitly tell them to do.
Autonomous AIs are here.
And -- tapping the sign -- this is another example where science fiction is becoming reality, and why “you watch too much sci-fi” is a stupid dunk.
This digital Westworld - “Smallville” - is now open source, meaning there will soon be thousands of these Westworlds, with thousands or millions of autonomous AIs populating them.
Makes you think, doesn’t it?
“Almost as interesting as the 7.8 billion inhabitants of Largeville, (almost) all unaware they are living in a simulation. Each has a unique personality and backstory.” - @Erikbryn
@DevaniAlkarim LA resident (and parent of young kids): would love data on correlation of homelessness density+poor performing schools and the correlation to non-SFR zoning.
Multi family wins IMO when our city defends livability as visciously as high-value SFR neighborhoods do.
"any sufficiently advanced technology is indistinguishable from magic" ✨ p sure i've never said this about anything we've made.
meet uncle rabbit—the first conversational holographic ai being powered by #ChatGPT@OpenAI powered by @LKGGlass
(hey @DBtodomundo 👋🏼)
Losing my mind about AI right now. I want to quit my job and work in the field while simultaneously vanishing into the forest to avoid thinking about the implications.
We're at the point where the exponential gains are starting. I can see it happening right now.
@lacontroller@kennethmejiaLA Yes! Can’t wait. And if LAPD wants to keep the military industrial state just swap the Whirlybirds for drones at 1/3 the cost and 99% safer.