Aaron Vanston (@aaronvanston) header
Aaron Vanston (@aaronvanston) profile picture

CTO & Co-Founder @buildpass_ - Building the world's smartest operating system for construction.

Melbourne, Victoria
Joined February 2015
Photo by Aaron Vanston (@aaronvanston)Photo by Aaron Vanston (@aaronvanston)Photo by Aaron Vanston (@aaronvanston)Photo by Aaron Vanston (@aaronvanston)Photo by Aaron Vanston (@aaronvanston)Photo by Aaron Vanston (@aaronvanston)Photo by Aaron Vanston (@aaronvanston)Photo by Aaron Vanston (@aaronvanston)Photo by Aaron Vanston (@aaronvanston)Photo by Aaron Vanston (@aaronvanston)
"Can we get SpaceX?" "We have SpaceX at home!" SpaceX at home:
Complete Tesla Roadster FLYING demo. INSANE!!!
3
124
Incredible how fast the languages developers learnt to code in are fading and general technical curiosity is what is being indexed on. I've never learnt Rust in my life, but have built some blazing fast rust ports of my TS libraries with the help of AI
2
38
Inspired by @theo forked version of ProxyCLI, I built and OS my own way to track agent usage and spend but also help manage my distributed set-up across all my machines Arbor puts all your Claude + Codex subscriptions behind one local proxy, smart switches accounts and manages machine health github.com/aaronvanston/arbo…
3
7
398
I trained a 36k-parameter model to centre a div. Stop asking trillion-parameter models. neural-flexbox: "flexbox, approximately" Inspired by the very useful gpu-lexer by @shuding . This is much less useful. I wanted to learn how to train a model from scratch. aaronvanston.github.io/neura…
I trained a small model to do syntax highlighting in the browser with GPU. Meet gpu-lexer from Vercel Labs: Small (27.5KB), fast (runs on WebGPU), and language-agnostic (model guesses the syntax). gpu-lexer.vercel.app It is experimental and built for learning!
1
1
2
12
2,296
Going fully AFK is the wrong goal for AI software factories. It's missing the point, and a lot of the value in what can actually be built with AI. Optimising for leverage and amplifying a product team's impact should be the main focus.
3
3
1
4
558
Aaron Vanston retweeted
Took me a while to find my place in the AI space. This year, I found it! Thanks all for recommending the courses. Helps me invest more time into skills, free resources, and interviewing badass people. More to come.
43
2
1
771
63,847
T3 Code remote connections just work so well, super easy to set-up, it just works... Well done @theo @jullerino and @t3dotchat team! Looking forward to the mobile app when its available!
2
160
Aaron Vanston retweeted
I’m gonna keep calling it codex because I can’t bear the shame of telling normies I use ChatGPT to code
91
114
23
4,182
128,133
Aaron Vanston retweeted
This is the relevant section of my CLAUDE.md I'll be real - I haven't read it much. Just vibed out what I was looking for with Fable, and had it confirm it can use Codex for the things I care about I still find Codex to be WAY better at computer use, verification of UI/UX work, and generally more efficient at execution on well spec'd work Minmaxing this has genuinely been really fun for me and I'm loving the outputs I've been getting. I was throwing away ~50% of my end-to-end agent-driven PRs before building this workflow. I haven't had to close a single one today :)
106
112
47
3,125
656,304
Aaron Vanston retweeted
I’m running a Codex meetup for @OpenAI Global Build Week Wed, 15th of July in Melbourne. A couple of short talks, then time to meet other people building with Al. Come along and bring a friend: luma.com/90uly47a
3
3
8
284
Anyone else having their codex app fully lockup constantly? I can see loading spinners, but cant click or interact with anything. (codex render CPU maxxing) Happening with long chats or firing off loops on top of existing sessions. Triggering every ~10mins, requires restart, and need to restart all sessions. Needing to move to CLI or third party app wrapper. Cant submit in app feedback due to the lock up, anything I could share to help unpack a fix cc @jxnlco @thsottiaux @ajambrosino
151
Anthropic servers heading on a well deserved break. Time to chuck the compute back on training?
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Claude models is not affected. We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible. Read our full statement: anthropic.com/news/fable-myt…
1
153
Built a tiny Chrome plugin for reviewing AI-generated PRs. I’m increasingly getting AI agents to attach screenshots, GIFs, and visual evidence to pull requests, but its hard to view them in GitHub quickly, cant zoom in and opens that media in new tabs. github.com/aaronvanston/gith…
1
3
102
In March, we shared The BuildPass Method with our whole team. It is our internal page for how we build, who we hire, and what we expect. Really, it is about a bigger question: what does it mean to build a company in the age of AI? method.bpass.team/method
1
2
6
133
We wrote the Method because expectations need a visible home beyond founder conversations and Slack threads. They should be durable and easy to revisit. If we say we are building for builders, we need to act like builders ourselves.
1
1
38
Sharing because more companies will need to make their operating model explicit. AI is changing the leverage of small teams and the pace at which companies can move. The companies that handle it well will rebuild around it with intent. method.bpass.team/method
1
39