product — flagship
a team of interns on your desktop
subintern is a native desktop app for running always-on AI agents — interns — you can give real work to. they plan, execute in a sandbox, and report back; you stay the operator.
the flagship subintern product, for macOS
most AI tools are a single chat you babysit. subintern is a roster: a set of named, persistent interns that hold their own context, run their own tools, and keep working between your messages. you delegate the way you would to a person — hand over the task, glance at the stream when you're curious, approve the risky steps, read the report.
what it does
- a roster, not a chatboxeach intern is a persistent agent with its own profile, memory, transcript, and tool permissions. give one the boring work, another the research, and check in when they report back.
- real work, really executedinterns run tools and plugins, execute commands in a sandboxed environment, and stream their thinking as they go — you see every step, and approve the ones that matter.
- your models, your keysan inference router lets each conversation run on the backend you choose — an OpenAI-compatible gateway, Claude Code, Codex, or OpenRouter — with local usage tracking per provider.
- native and unobtrusivea polished macOS app that sits beside your work. interns message you when something needs you; otherwise they just get on with it.
subintern for desktop is in active development — we're preparing a first early-access build for macOS.