THIS WEEK’S STORY
“Vibe Coding” Is Shipping Real Infrastructure
A few days ago I found myself doing the kind of work that usually scares people away from shipping.
The “unsexy, production-grade stuff” that actually determines whether an app survives real users.
👉 Securing every API route. Server components. Input validation and sanitization. Rate limits. Bot prevention. LLM caching and optimization. Bulk requests…
Two days later, it was done.
Not hacked together. Not duct-taped. Secure enough for production, faster, more scalable, and cheaper to run.
Overall, LLM costs dropped and loading times improved. All of it built with AI, end to end, for a few bucks in credits.
This is the part people still don’t want to accept.
The bottleneck isn’t technical execution anymore. It’s knowing what needs to be done, why it matters, and how the pieces fit together. Once you can reason about the system, AI does the heavy lifting with you, not instead of you.
I saw someone on X sum this up perfectly after being told that “vibe coding isn’t real engineering.”
They shared how a two-hour CDN overhaul cut bandwidth by 99%, dropped costs from $12K to $200 per month at scale, and required zero backend changes. While others were debating definitions, they were shipping real improvements that move the business forward.
That’s the shift.
Engineering used to be about who could write the most code. Now it’s about who can make the best decisions, the fastest. Security, performance, cost, UX, tradeoffs… AI turns those decisions into working systems almost immediately.
If you’re waiting for permission to build because you don’t feel “technical enough,” you’re already behind.
The people shipping right now aren’t louder or smarter. They’re just building while everyone else is still arguing about what counts.
Go build 🤘
🤖 PS: if you wanna do it faster, join this waiting list (few spots available)
HEADLINES
Lovable Snags $330M Funding for Vibe Coding Revolution
Swedish startup Lovable just raised $330M at a $6.6B valuation, with backing from Nvidia and Google Ventures. Their pitch is simple; build full apps using natural language instead of traditional code.
This matters because it confirms something important. “Vibe coding” isn’t a side experiment anymore. Serious capital is flowing into tools that turn ideas, prompts, and product thinking into real software. For non-dev builders, this means you can prototype fast, validate ideas cheaply, and only go deeper when it actually makes sense.
Google's Gemini Gets Vibe Coding Boost with Opal Integration
Google quietly shipped Opal inside Gemini, a no-code way to create small AI apps by combining drag-and-drop logic with AI-powered text and image generation.
The signal here isn’t the UI. It’s the direction. Big platforms are betting that building software should start with intent and outcomes, not syntax. For non-developers, this means spinning up internal tools, workflows, and prototypes in minutes, and spending time on what matters most; solving real problems instead of wrestling with setup.
THIS WEEK’S TIP
Let Claude Code do your app security
Making sure your “vibe coded” app is secure, stable, and hacker-proof, is essential if you want to launch real applications even as a non developer.
The goal is to let AI do the heavy lifting but you need to point it in the right direction. So here’s what you need to be aware of:
API routes security → are api routes on the server-side and not exposed on the client-side, meaning anyone can view them and manipulate them?
Hard-coded API keys → api keys should be securely stored in the .env or server-side (Vercel) fro peoduction
Input validation & sanitization → are you checking if the input (length and type) being typed in by the user in forms, AI features, and across the app is valid?
Rate limiting → are you limiting the amount of requests per minute your users can do?
No auth on internal endpoints → make sure all endpoints have authentication protection
Debug code logged to console → do you have debug snippets showing in the console?
And here’s what you should do next:
Go to GrokAI and ask it to give you the most important things to do to secure your nextjs, react native, or other application, and give the examples above
Open Cursor and open the Claude Code extension (you need to install it first)
Type /init so the agent creates an overview of your codebase in a CLAUDE.md file
Ask it to run a deep security assessment of the project considering the best practices detailed by GrokAI, and instruct it to write everything up split in phases in a new security-assessment.md doc
Open a Cursor chat with GPT5.1-Codex-Max in PLAN mode, and ask to read the security-assessment.md doc and create a detailed implementation plan for phase 1
Sit back and relax – it will start implementing the changes and instruct you what to do on your end (setting up external services like Upstash Redis and QStash…)
IMPORTANT: Commit and push the changes to your GitHub repository
Move on to phase 2 and repeat the process from step 5-7
And that’s it, you just made your app secure in a few hours.
This is what smart engineers are doing right now with AI coding and you should do it too.
🤖 If you need any help or guidance, join this waiting list (for a small group only) ✌️
QUICK HITS
Anthropic acquires Bun to supercharge its Claude Code agent, which has already surpassed $1 billion in revenue just months after launch while dominating coding benchmarks
OpenAI releases GPT-5.2 with a strong enterprise focus, featuring enhanced multi-step reasoning, long-context handling up to 256k tokens, and superior performance in professional tasks like coding and document analysis
Nvidia unveils the Nemotron 3 series of models, optimized for agentic AI with massive context windows up to 1M tokens and high efficiency for multi-agent systems and autonomous applications
Google launches Gemini 3 Pro, powering advanced multimodal agents including upgraded Deep Research for complex queries and faster, more practical generative features like real-time image and video creation
AI coding tools drive massive productivity gains with 65% of developers using them weekly, though a Stanford study shows a nearly 20% drop in entry-level software jobs for young developers since 2022
PS: I’m working on practical guides and a small live cohort to help non-developers actually ship real apps with AI (not just learn theory).
I’ll be opening this to a small group first,
join the waitlist here → https://tally.so/r/gDeLrM
Hope you enjoyed this aiOS App Builders edition!
💡 Got an AI tool for us to check out or collaborate?
Send us a message and let us know!
Was this edition forwarded to you? Sign up here
See you next Saturday!
Filippo
