Rust's error messages are still unrivaled.

CUDA, CUDA, CUDA... What if there was a different, open way to write highly-efficient GPU kernels? This is the big swing that Tracel AI is taking. And, guess what... Rust is key to their approach. Check out our podcast with Tracel CEO Nathaniel Simard (linked below):

What are things that software engineers think are really important that just aren't when you zoom out?

Is there such a thing as artisanal software? If so, what does it looks like? Examples? Artisanal: products made by hand or using traditional, non-mechanized methods by a skilled craftsperson, typically in small batches

If LLMs are good at translating from one programming language to another (I think they are), then we should be seeing less programming language-specific requirements in software engineering job postings. For those in the job hunt, does this match what you're seeing?

Seems like we need to figure out continual learning for the next step of AI... The fact that SOTA is models built in Python is a big barrier to getting this done since Python is not highly cross-platform. Does anyone know of a language that might be good for this...? 🦀😉

I've been really impressed with Agent Skills as a way to extend the functionality of coding agents. So, I had a thought... A community led and maintained skill for writing Rust. 🧵

I'm very intrigued by OpenCode for one reason: Ability to switch models quickly without relearning the whole tool. For those that are using it, has they delivered on this value prop? Or, are you still reworking your workflow when swapping models?

Rust is exploding in the defense space. Is that just because of expanding defense budgets? What other factors are at play here?

Are you guys writing specs for your agents to code from? How are you doing it? - I assume it's markdown - Is it multiple files? If so, what is each file for? - Is there a standard name you always give the file? Or a naming convention? - Where do the files live?

If you are primarily editing files with a CLI-based agent (Claude Code, Codex, etc.), how are you reviewing the files? In an editor or just right in the terminal with vim or something else lightweight?

Why do people like Zig over Rust? Is it more of a reaction to a certain culture that exists around Rust than to the technology itself?

Most commentary about the usefulness of coding agents seems to center around web, mobile, etc... Are the agents similarly effective in real world domains? (embedded, robotics, aerospace, etc.) If not, why not? Please repost for reach and @ mention those who might have insight!

You might want to unfollow us... Our content has always been very Rust focused. We love Rust. Beyond the language, we love the Rust community. And, our highest purpose has always been to help those in the community find ways to support their families through rewarding work. 🧵

AI Code Review Is A New, Discrete Skill We have to learn to discern when we have to understand a snippet of AI generated code deeply and when we don't. - Bit of UI code that behaves how it should? -> Skim - Password salt and hash-> CHECK IT Agree?

Elon said Rust is pretty great. Rust people on X:

I always thought the Rust haters were grizzled C/C++ vets who couldn't be bothered changing because they had expelled memory leaks out of their very being many years ago. Now I'm starting to see that a lot of them are just as zealous as the RIIR crowd.

Help me write questions for our next podcast interview! What questions or thoughts do you have about the intersection of AI and Rust?

Why is Rust such a big deal for robotics? Carter Schultz of @ampsortation.bsky.social explained this to me in a very clear way when we talked. Check out the interview linked in the comments!

January '26 Rust Jobs Report just dropped! Highlights: - Privacy company Proton cracks top 10 hirers - IOT/Robotics/Automotive finally takes its rightful place as largest Rust hiring industry Check out the full report linked in the comments!

Ignore every AI headline. Many of us are breathlessly following every development. It's keeping us from the daily blocking and tackling that actually moves the needle in our lives. Learn the state of the art for a couple of concentrated days every quarter. That's all you need.

If you write code in 2026, it seems there are two ways forward for your career: 1. Learn how to build a whole product (learn about design and business). 2. Learn even more about the deep tech details. Which way are you going?

Are you Rustmaxxing?

Been working on learning how to wield the term "mog." So, let's give this a shot... Fair to say Rust mogs C?

STOP SPAMMING YOUR RESUME TO RANDOM JOBS! Instead, think like the hirer. Q: What is a hirer's greatest fear? A: Hiring a poor fit, or worse, someone dishonest. To get hired, think about providing off-resume evidence of capability and character. Here are some ideas: 🧵

@rustcodepro.bsky.social and @filtra.bsky.social -> you might be interested in the interview we did with Nathaniel Simard, creator of Burn, on @practicalai.bsky.social practicalai.fm/242

If you DON'T code for several hours every day, you probably like coding with AI. I'm looking at you C-Suite types and podcasters. (Also me to be fair) If you DO code a lot, you probably don't like using AI. That seems to be the way this debate is breaking.

The whole git adopting Rust thing seemed to bring out an ugly side of conversations about the Rust community. No idea why. But, some unsolicited thoughts: 🧵

I'm sorry is there suddenly a movement to say that Rust is not extremely cross-platform? Where is that coming from?

PCs, mobile, AI... What do these have in common? All technology waves that IBM arguably "missed" One wave they won't be missing is quantum! In our recent interview about Rust use at IBM Quantum (linked below), Matthew Treinish explained how IBM got into quantum:

New Rule: New projects must be written in Rust. *This is specifically about projects requiring compiled languages. *Also, know that there are lots of reasons why your specific project might require a different language. You'll know if it does. If it doesn't, just use Rust.

Aged well.

How important has reading code been to your process of learning software engineering?

Has anyone tried having AI write machine code? How does it do?

Literally what Rust is doing for Python these days 🐍🦀 (photo credit Allen Pan)

Rust continues its march to power all of the Python ecosystem... Samuel Colvin of @pydantic.dev just announced Monty, "a minimal, secure Python interpreter written in Rust for use by AI" Learn more about the 🐍 + 🦀 magic happening at Pydantic with our podcast linked below:

When a non-engineer asks if I'm worried about AI because it's "good at coding," and I ask what happens to their job when AI can code anything.

If AI is good at writing Rust, is the value of humans that write Rust well going up or down? Arguments For Up: - AI code still needs review Arguments For Down: - It's easier to review code than to write it - AI makes it easier to learn Rust

Does AI write more memory safety bugs than humans?

Rust is the real deal. Jeff Vander Stoep posted the receipts in his blog post about Rust in Android: - Memory safety vulnerabilities: 70% → 20% - Rollback rate: 4x lower than C/C++ - Review time: 25% faster than C/C++

Is Rust actually slow to write? Not if you’re used to C++. Most people who call Rust "highly productive" are looking at it next to older systems languages. In that world, Rust wins every single time.

What's the feature set for a coding agent that specializes in Rust?

People are claiming to be generating 10k LOC per day. What.

Been thinking about how to scale the Rust ecosystem. I think the simplest path is to buy ChatGPT ads against all code queries. Ideally, you could basically tell ChatGPT to write Rust instead of any other language. We’ll see what's possible. 🦀🚀😏

You do not need a physics PhD to get into quantum computing. In our recent interview, Matthew Treinish shares how he moved from open-source software into quantum (link in comments).

Is it time to fight AI with AI? Anyone working on the PR sorting agent?

Rust code that compiles ≈ correct code. Not enough people know this.

So is there anything that could be done to make linking to non-Rust libraries from Rust less painful?

Rust has a linear algebra problem. See, linear algebra is super important for many of the applications where Rust is most commonly used. Yet, we have an underwhelming set of linear algebra libraries. Matthew Treinish of IBM Quantum explains why they're underwhelming 👇

My therapist says that writing custom types in Rust is as effective as meditation.