@leveragecpu hiring: https://xitter.cf/t.co/lrEH7bE84u
San Francisco
Joined October 2024
- Tweets1.2K
- Following91
- Followers939
- Likes1.2K
I use this a lot
Anyone else?
When working with agents, I'm usually a few steps ahead. I read their thoughts or come up with other things I want them to address while they're still working.
Some things can be kicked off in parallel, but often I want to steer the agent in a specific direction and build on its existing work.
Right now, I capture these thoughts in a note or scratchpad on my desktop.
Having a way to stash them directly in the agent UI and bring them back when needed is really useful. We shipped this in @leveragecpu . It also syncs across devices.
this is really cool
Made a skill for dimensional drawings of UI components.
Inspired by Apple's accessory drawings, it documents a component's architecture into a spec sheet.
github.com/enesozturk/dimens…
Working on a new desktop experience for @leveragecpu.
Gen UI is powered by @typesafeai's Jev.
.@leveragecpu now also uses Pele for visualizations. Video created by Leverage
One week later, Pele is already in dozens of apps including @obsdmd 1.15 and @bot! Many improvements this week:
- Even faster rendering!
- Math resolver support: KaTeX, MathJax, Temml, etc
- Gradients in sankey diagrams
- Fewer edge crossings in diagrams with subgraphs
- Subscript, superscript and inline code in labels + text styles: kbd, underline, ins, cite and others
On a personal note, I am so grateful for all the love Pele received! I truly did not expect it.
My hope with Pele was to get a few bug reports ahead of the Obsidian release, so it's a wonderful feeling to see it adopted so quickly. I take this as a great honor and responsibility.
If you added Pele to your app or website, feel free to share it below 🌋
I've read/scanned thousands of applications in the last three months. The best ones from my POV are either extremely short, as in max 3 sentences or a few bullets ( eg. Hi, I'm good bc 1,2,3 <- has to be actually something relevant and ideally impressive. can be a demo as well) or very personal, handwritten notes that show something special in terms of your fit. The worst applications are always long walls of LLM slop with 10 links to vibe coded projects.
Other founder friends I've spoken to share this sentiment. So, sharing this as anecdotal advice.
Legibility is a precondition for delegation. It held for provincial administration and for the joint-stock company, and it holds for agents. You cannot hand off what you cannot read.
🔍
When our AI code reviewer's cost per run doubled, we dug into the token usage to find out why. Learn how we used prompt caching and intelligent model routing to slash our inference bill by 68% while speeding up reviews. cfl.re/4isj2NU
Earlier we tested new notification sounds for Leverage Computer. The whole thing started in one thread that @io_gabx started. @0xRaduan then prompted Leverage and the agent built an artifact with the candidate sounds, inline, and everyone tried them out inside the thread. When someone wanted to change something, they either prompt in the thread or open the session view to go deeper (thinking traces, raw files etc.).
The artifact itself can be shared internally or externally, so a small design decision like this does not need a separate tool or a meeting. It is a minor example, but it is roughly what AI-native work can look like in practice. Start where the team already talks, and go as deep as the task requires without leaving the workspace.
The @leveragecpu web and desktop apps are live. Worth trying imo.
