Rally
Notes from the studio. How projects get scoped and priced, what shipping to the App Store actually involves, and the numbers behind our own products. Written for people who have to make these decisions.
Just the good stuff. No spam.
Why we stopped using tmux for AI coding agents
We ran Claude Code and Codex inside tmux for eight days. Dead panes and blank reattaches pushed us to a small Rust daemon that keeps every agent's screen.
Read the post ↗
What is on-device AI, and does it actually keep your data private?
On-device AI runs on your phone's own chip, so what you type or photograph never leaves it. Since iOS 27, "built with Apple's AI" no longer guarantees that. How to tell, and how to choose.
How to open source a side project (what Agency taught me)
Before you open source a side project you built for yourself, run it on a machine that has never seen your setup. What open sourcing Agency took, from logs and signing to a Linux port.
How to run multiple AI coding agents in parallel without them overwriting each other
Give every agent its own git worktree and branch, then review each diff before you merge. The steps by hand, what Claude Code, Codex and Cursor automate, and the traps.
How to use Apple Foundation Models without them making things up
Apple's on-device model is good with words and bad with facts. In Senba, code picks the facts, the model writes the words, and a validator checks every line.
Why every dumb phone app for iPhone looks the same
Apple rejects home screen replacements, so every dumb phone app for iPhone is a text-list widget that opens Spotify. What a real one has to own instead.
How to get a fixed-price quote for an app (and what it should include)
A fixed-price app quote that holds lists every screen and state, prices each line, adds a change budget, and moves anything unknown into paid discovery.