Engineering write-ups, product updates on Romeo, and notes on AI and agentic development — search or filter by category below.
Why Romeo runs as a desktop app, not a browser tab
The tradeoffs behind giving an AI assistant real access to your machine instead of sandboxing it in a browser.
Sep 2026 · Read article →Agentic development, explained for people who ship things
What "agentic" actually means in practice, and where it breaks down once real users show up.
Our build pipeline: one team, three product lanes
How software, mobile, and digital assets share the same underlying systems instead of forking into separate stacks.
What we scope before writing a line of code
The questions that decide whether a client project becomes a fixed-quote build or an ongoing engineering partnership.
Letting an agent touch production: our guardrails
The permissions model behind Romeo, and why we still keep a human in the loop for anything irreversible.
Picking a model is not the hard part
Notes on evaluation, latency budgets, and fallback behavior when the AI layer of a product goes down.
Founding the studio: what changed after year one
Independently owned, no outside investors — what that's actually meant for how we pick projects.