Pinned Tweet
Atriana is a CAD tool I'm building where the agent does the CAD. I wrote the three things it needs to be useful: a way to draw, a way to see what it drew, and a way to show the result to a human. Here's Claude modeling a wood-burning stove.
🎯
There’ll be times where you over-guide the AI. It knows better than you and is more intelligent than you.
There’ll be times where you under-guide the AI. It can go off-track, have the wrong biases, or miss context.
How well you manage this superpower is the #0 skill I look for.
Tomas Gonzalez retweeted
There’ll be times where you over-guide the AI. It knows better than you and is more intelligent than you.
There’ll be times where you under-guide the AI. It can go off-track, have the wrong biases, or miss context.
How well you manage this superpower is the #0 skill I look for.
Tomas Gonzalez retweeted
Tensordyne gave Aristotle its hardest block to verify: 12.2 million gate equivalents. 185 properties proved, 56 refuted.
Jonathan Ko, their verification engineer, calls it "unthinkable with traditional tools."
Read the the full @TensordyneInc case study: harmonic.fun/aristotle/case-…
Tomas Gonzalez retweeted
Excited to partner with and support @tannerlinsley and team, champion of React, open source, and the web
What this means for our customers: whether you choose @nextjs or @tan_stack, we are committed to the highest quality service and support.
Tomas Gonzalez retweeted
Beyond cool to see @Vercel as our latest partner! TanStack-everything has always worked well on Vercel, so making it official leaves no doubt that it's going to get even better. 🤝🎉
🏝️ + ▲ = 🚀
TanStack and @Vercel are officially teaming up!
We’re working together to make Vercel a first-class home for TanStack Start with better support for deploying and running anything you can dream up with TanStack on Vercel.
Start here! 👇
vercel.com/kb/guide/deploy-a…
Tomas Gonzalez retweeted
used a trick @jlongster came up with
agents can control browsers but you can also ask it to record network requests into a HAR file
then it can derive a client for any website which is more efficient than browser controlling it every time
made it build a quick uber eats cli
Tomas Gonzalez retweeted
Learn to ship. Shipping is a skill distinct from coding. Shipping is designing, coding, QAing, story-telling, teaching, marketing, selling, pivoting, iterating…
It used to be that coding dominated in importance because of coding ability scarcity. AI will push you to go further.
It runs with any MCP client. The app also sends screenshots back to the agent so it can review its own work and iterate.
gztomas.dev/projects/atriana…
Don't ever, for any reason, do anything, to anyone, for any reason, ever, no matter what, no matter where, or who, or who you are with, or where you are going, or where you've been, ever, for any reason whatsoever. Never try to start the dev server, it's already running.
There's something circular about testing: you can only write tests for bugs you've already imagined. The ones that reach production are always the ones you hadn't considered.
I've been trying something different. Instead of test scripts, I write product specs in markdown. Plain descriptions of what each feature does, from each persona's perspective, with invariants the system should hold. Then an AI agent reads the spec and tests the app the way a user would, probing edge cases I never thought to script.
It's not a CI replacement. It catches a different category of problems: the bugs that live in the gaps between features you planned. Wrote about the full setup and what it actually finds.
gztomas.dev/blog/specs-the-t…
Made a tiny CLI so Claude Code can talk to me using ElevenLabs TTS.
Just tell Claude in your CLAUDE.md to use utter me for responses. That's it.
github.com/gztomas/utter
Tomas Gonzalez retweeted
The new software development paradigm: the purity of the codebase is no longer the holy cow protected by angry senior devs.
Companies will use agentic code systems like Claude Code Opus 4.5 to fix messy codebases as long as they are fixable.
Once each new fix starts taking too much time or consuming too many tokens, the entire non-fixable piece will be rewritten from scratch based on the spec and the unit tests remaining from the previous non-fixable version.
No human will be involved in the code writing. Only in the spec and test writing.