Today we’re open-sourcing our own version of “Claude Design.”
It uses your local coding agent to do the work Claude Design can do.
We also have a cloud version currently in private beta. Stay tuned.
https://t.co/x5wEteUuQ5
Rebyte now supports bringing your own API key and subscription (available for Pro users). We’re committed to making pricing more transparent by showing a detailed credit breakdown for every run.
With subscriptions, we run the exact same Claude Code and Gemini agents you’d run on your local computer.
We’re pleased to announce Agent Computer Cloud: like a traditional computer, but the only interface is through an agent harness. It persists forever, runs only when you interact with it, and lets you own an unlimited number of agent computers.
Same idea as claude code routine
Rebyte routines let you turn a prompt into something that runs on a schedule and sends back useful results. back by claude code/codex/gemini cli/opencode, you define the agent.
I just posted a demo of a routine that watches US market hours, checks a few stocks every hour, compares prices vs. the previous hour, analyzes what changed, and sends the summary back automatically.
You can use the same idea for anything you want to monitor:
stock prices, watchlists, company news, competitors, filings, crypto, niche indicators, whatever matters to you.
Set the routine once. Rebyte keeps watching. You get the updates.
That’s the point of routines: not just asking AI once, but having it do recurring work for you.
#buildinpublic
Just shipped: build exceptionally beautiful slides with coding agents 🤩
Describe what you want → agent writes the code → polished deck, live instantly.
No more fighting PowerPoint. No more bland templates.
Try it → https://t.co/GlETopq5uK
#buildinpublic
Do you really need "Data agent"?
The tools for data work already exist:
DuckDB, Python, SQL, Jupyter, charting libraries.
What was missing was usability.
Now a general coding agent can use those same primitives for you:
import data, analyze it, and turn it into a report — from plain English.
Less structure, more tools.
https://t.co/asQfbnHXt3
Unlike Anthropic’s managed agents, we choose to use the sandbox as the total environment for the agent rather than separating the two. The benefit is obvious: it is supremely easy to debug and reason about because everything lives within a single, persistent file system.
Beyond just ease of use, our approach is open to all agents. You can run Cody, Claude Code, OpenCode, or the Gemini CLI all within the same space. While we plan to add agent orchestration later, providing a unified environment allows different models to work effectively on the same codebase today. We aren’t against Anthropic’s managed approach; we are simply solving a different set of problems.
Launching https://t.co/Q3s7zK0qeK today.
We built professional office skills for AI agents, so they can create, edit, and refine documents, Excel files, and PowerPoints end to end.
No manual formatting grind. Just agents doing real office work.
A Midjourney engineer just open-sourced a way to lay out entire web pages without CSS.
Not a framework. Not a library that wraps the DOM. A pure TypeScript text measurement algorithm that bypasses browser reflow entirely. 600x faster.
Why it matters: AI-generated interfaces need to lay out text dynamically. The browser wasn't built for that. CSS reflow was designed for humans writing static pages, not agents generating UIs on the fly.
This exists because agents need rendering that doesn't depend on a 30-year-old pipeline.
The AI-era stack is being rebuilt from scratch. One library at a time.
P.S. Such a fun to play with that!