LLMs fail at raw SVGs: wasted tokens & prompt injection risks.
My fix for Thousands Blade: Dynamic SVG Injection ⚡️
1️⃣ Visually select icon 2️⃣ Saves direct to Assets folder 3️⃣ AI imports SVGs as a component, never reading the XML
🛡️ Secure 📉 Reduce Token Usage 🎨 Easy stylable
@madhavjha LLMs fail at raw SVGs: wasted tokens & prompt injection risks
My fix for Browser AI-IDEs: Dynamic SVG Injection ⚡️
1️⃣ Visually select icon 2️⃣ Saves direct to Assets Folder 3️⃣ AI imports SVGs as a component, never reading the XML
🛡️ Secure 📉 reduce token usage 🎨 Easy stylable
@yyjhao Hey Yujian—saw Floot’s UI→AI edit idea. I built a small prototype where AI edits live React components in the browser.
Quick demo (6 min):
https://t.co/kDXE95Tc7R
Nik has been on a roll, and for someone who started ~50 episodes ago and does it as a side gig, he has really improved over time. Full pro-podcaster he has become. 😀
Goes on to show that the odds of success are much higher when you build things around what you love.
@nikhilkamathcio is curious, looks good, listens, and speaks well, and the magic sauce is launching the podcast at the right time when podcasting in India was just taking off.
The thing that stood out to me was what @elonmusk said beautifully towards the end of the podcast, and also describes much of what Nik does with WTF, which I’m proud of as a brother.
“I’m a big fan of anyone who wants to build. Anyone who aims to make more than they take has my respect. That’s really the core principle: strive to be a net contributor to society.
It’s similar to the pursuit of happiness. If you want to create something valuable or financially successful, the goal shouldn’t be the money itself. Instead, focus on providing genuinely useful products and services. When you do that, the financial rewards tend to follow naturally. Just like happiness—you don’t chase it directly; you pursue the things that lead to it: fulfilling work, learning, friendships, loved ones. The happiness is a byproduct.
It might sound obvious, but anyone starting a company should expect to work extremely hard, accept that there’s a real chance of failure, and stay focused on ensuring that the output is worth more than the input.
Are you creating value? That’s what matters. Make more than you take.”
I have sought relief from the SUPREME COURT OF INDIA through a HABEAS CORPUS petition against @Wangchuk66’s detention.
It is one week today. Still I have no information about Sonam Wangchuk’s health, the condition he is in nor the grounds of detention.
Blunt truth: in 10 years, the engineers left standing will be the ones who:
Supervise & guide AI
Architect complex systems
Secure AI pipelines
Connect tech to business
AI eats average. The top 5% will run the table. Play for that.
AI now writes code.
If you’re a software engineer, the next 10 years won’t look like the last 10.
The number of jobs will shrink, competition will spike, and AI will eat junior work.
Here’s how to survive (and dominate) in the AI era 🧵👇
5. Move up the value chain.
AI automates coding tasks. It can’t replace product thinking.
Learn to connect tech → business outcomes:
saving costs
speeding delivery
enabling new revenue streams
That’s where promotions, startups & $$$ live.
4. Build a reputation, not just a resume.
The global talent pool is huge. Your personal brand is your moat.
Publish open-source projects
Share insights (“what AI got wrong & how I fixed it”)
Carve a niche: e.g. “engineer who makes AI-human systems reliable”