For 2026:
I hope you find the courage to build the strange ideas that keep you up at night, and the ones you’d regret not creating.
Build things that matter, break things to see how they work, and ignore the hype cycle.
The grifters will fade when the algorithm changes. You remain.
Take a breath. Close the laptop. Hug your people.
I love the potential of AI, but it also makes me a bit sad. Vibe coding feels like undoing years of hard-earned craftsmanship and precision. The challenge now is using AI to raise the bar, not lower it.
It sounds crazy doesn't it?
Till you read the fine-print
> The compiler generates less efficient code than GCC with all optimizations disabled
> It doesn’t have its own assembler or linker
> Can’t produce a 16-bit x86 code generator
> it has “nearly reached the limits of Opus’s abilities.”
not to mention, the had the GCC compiler in the feedback loop, and even within training data I'm sure a TON of GCC compiler code already exists so Opus has seen what it's writing
Hype aside this is actually a big deal, a senior compiler engineer costs a ton of money and Opus did the bulk of the work for much cheaper and it will only get cheaper
Excited to see what we have by the end of the year if the start is so good
@nachocodoner@meteorjs Oh, this is huge. 🚀
It’s ironic that years ago the industry rushed to React, but today Blaze feels like the more stable, 'get-stuff-done' option while React keeps reinventing itself.
Huge props to the community & team for keeping it viable in 2026.
@trishlaostwal Efficiency gains here look huge, but the security implications are terrifying.
we are basically giving ai-agents a standardized API to spend money before we've figured out how to stop them from falling for prompt injections.
For years, the #1 Meteor complaint: "bundles are too big."
Fair criticism. No tree shaking meant dead code shipped to users.
Meteor 3.4 fixes it.
Our test app (simpletasks): 50% smaller bundles.
Tree shaking is finally here.
@simonw I'm currently stuck between L2 and L3, and honestly, the cognitive load of verification is the biggest wall.
It’s hard to imagine L4 being feasible when simply reviewing the output of a L3 workflow is already this mentally expensive.
L5 sounds fascinating but distant.
i'm so excited to be building @witty_social. i'm at $20 MRR while competitors in the space are at $20K MRR (and higher). this doesn't put me off, this tells me the area is massively validated and people want growth tools for 𝕏.
Tada! 🎉 I'm so excited
My Pineapple Log app is officially live and just got approved!
Please go to the App Store & search for Pineapple Log
It is a diary app, the ease of scrolling through a timeline, the feel of social posting to journaling - without the performance.
Capture your day in whatever format feels right.
It’s completely free right now, so please download it, try it out & send me some feedback! 🙏
You vibecode all day but still track time like it's 2015.
Timers you forget to start.
Spreadsheets you never update.
Invoices you guess.
I built https://t.co/1NB39MTHbs
Press a button. Talk. Time tracked.
That's it.
Announcing the first preview release of my new server management & app deployment platform built entirely with SvelteKit.
It will soon let you connect to various cloud providers, or add your own servers on-prem.
If you like what I'm working on, please Sponsor or Donate to this project. Your contribution is greatly appreciated. Lord knows it's been a rough last couple of years.
My goal is to build great open source software that people will love to use.
https://t.co/pPyP0e7bhx
@burkov Watching the Head of HR cry about 'company values' the moment he finds out his own department is being restructured.
Bestie, you wrote the severance policy!