Bitrise Remote Dev Environments

• Высокие операционные расходы на предоставление и масштабирование облачных macOS‑инстансов из‑за ограниченного предложения Apple‑hardware.
• Конкуренция со стороны крупных облачных провайдеров (AWS, Azure, GCP), которые могут быстро предложить аналогичные услуги с более низкой стоимостью.
• Зависимость от экосистемы Bitrise: если пользователи перейдут на другие CI‑системы, ценность RDE резко снизится.

“ Hey Product Hunt, I’m Arpad, VP of Engineering at Bitrise (mobile DevOps startup). Agents need somewhere real to run. You can hand an agent a task , but it has to execute somewhere . ❌ On your laptop it fights you for the CPU and the simulator, running multiple in parallel and you won’t be able to do a video call without CPU clipping and fans screaming. ❌ In a throwaway container it has no real device tooling, no warm dependencies and no memory of the last run, so it spends its first tokens rebuilding an environment it’s about to lose again. ✅ Bitrise Remote Dev Environments are that somewhere: Bitrise RDE are on-demand cloud VMs (real macOS on Apple Silicon, or Linux ) that an agent drives headless through our open-source MCP server. Start a batch on different branches, → let them work in parallel, → collect the PRs. Or attach VS Code, Cursor or Claude Code over SSH and work in it yourself, like a laptop, but many. The reason it actually works: these are the same machines that run production builds for mobile teams at scale. Same Apple Silicon hardware, same stacks, cloned from the same golden image the CI build fleet uses. 🤷‍♀️ Honestly this came out of a thing that annoys all of us. CI fails on something your laptop never hits: a different Xcode, a stale pod cache, a simulator runtime you don't have locally. You pull the branch, it passes, and your afternoon is gone comparing two machines instead of fixing the bug. Agents made it worse: same problem, none of the judgement. It burns tokens down rabbit holes, sometimes fixes the environment only to hit the same wall next run in a fresh VM, then confidently opens a PR that can’t build – or fixes the wrong thing. So we gave the human and the AI agent the same machine. When the environment you write code in is the one that tests it, a whole category of works-on-my-machine stops existing. Ask me anything about how it works, what it costs, or where it still falls short. ”