
Building @OpacityHQ (now Zoah). Previously @GitHub. MSc Software Engineering @UniofOxford. https://teapawt.net/t.co/6UVlXpFRHY
Joined May 2010
- Tweets2.8K
- Following3K
- Followers767
- Likes6.6K
Replying to @steveruizok
Totally agree! I love how easy it is to use LLM’s for performance work. Removes the toil of benchmarking and profiling, and makes it easy to create a custom harness and visualization of results. Love subagents to explore hypotheses in parallel. Divide and conquer all the way down
4
390
Recorded a 12-minute end-to-end tour of @parkwebhq last night, now that it's stable enough to use as my primary browser. Includes a few surprises I haven't talked about yet, like shared spaces, workshop features, and stacks. Park is evolving into something much more collaborative you can use with friends, family, and teams 👌
Replying to @kostyafarber_
I’m not a huge fan of cyclomatic complexity as a measure of code quality in general, but for agent code I think it’s a good signal to identify missing abstractions, or implicit state machines.
1
1
52
Replying to @kostyafarber_
I’d love that!! Some other ideas I was imagining are cataloging core call paths in the app, annotated with performance metrics that update based on CI runs (so you can diff perf), using similarity to identify potential duplicate code candidates. And viewing call paths by module.
1
1
34
Replying to @kostyafarber_
Awesome! Yeah Pierre diffs and trees are so nice! I have some ideas in this space, maybe we can swap notes sometime? Excited to see what you come up with!
1
1
25
Replying to @kostyafarber_
This is super cool! What’s the editor you’re using? It looks like vim mode?
3
1
24
Empirically I've seen that by specifying a solution in terms of its structure, my coding agent produces better results (quality / time / tokens) than narrative descriptions alone.
This study measures that gain and provides its own metrics for evaluation.
arxiv.org/abs/2610.38214
1
2
166
And don’t forget Go concurrency patterns from 2014!
go.dev/blog/pipelines
i was reminded of this blog from 2016 on visualizing concurrency in go today. i might start having opus visualize my programs this way
divan.dev/posts/go_concurren…
4
244
Every browser I've ever used has the same developer tools we all struggle with as designers. I wanted to flip that script and build a comprehensive set of designer tools in @parkwebhq. Just a tease, but you get the idea. Join the waitlist parkweb.co
Replying to @sliminality
The use of a sheetrock anchor to represent ⊥ is top tier (although it is bottom)👏
3
303
Lovely! I love the animation from application to normal form.
Dave Keenan's 1996 lambda calculus visualizations, finally real.
dkeenan.com/Lambda/index.htm
claude.ai/artifact/DfG2y96ST…
2
131
Incredible! Imagine a world where all critical software was formally verified. Or a world where formally verifying your systems was as easy as a prompt.
We are out of stealth! As AI writes more code, we're building tools to prove programs meet precise requirements, with Lean proofs anyone can independently check. Let's make verified software the norm in every language and application.
Come take a look: andtruth.com
90
rewinfrey retweeted
As we get deeper into the age of AI, the design process has started to feel muddy. We're fixing that.
In Zoah, you build an identity first, and it becomes the foundation your product is built on. Here's a sneak peek. 👀
rewinfrey retweeted
Introducing Zoah MCP.
Claude Code, Codex or any MCP agent can now open a Zoah project and design with its real components and variables.
I've become super passionate about delivering the best brand builder I possibly can in @OpacityHQ. Instead of prompt to brand, I've implemented a process of deep research that develops an identity, only after which we can begin the visual journey of creating a solid brand.













