Skip to content
Remote Desk
Built in — not an integration

A full AI workspace, right at your desk

Chat, compare models side by side, run deeper research, and generate images — all from one surface in the same office your team works in. You don't pick a routing mode or a model: the system reads the task and chooses both, so you write the prompt and get the right answer.

Start freeBook a demo
COMPARE · 2 MODELSYOUDraft a launch announcement for the new release.Model A · fastUSE THISModel B · deepUSE THISAsk anything, or refine a response…
The flowMeetCaptureAct

One surface, many capabilities

Chat, compare, research, and image — in one place

Ask a quick question, compare how different models answer, run a research pass over a harder question, or generate an image — without switching apps or tabs. It's a single workspace that flexes to the job in front of you.

  • Multi-model chat with full conversation history
  • Side-by-side model comparison
  • Research mode and image generation built in
COMPARE · 2 MODELSYOUDraft a launch announcement for the new release.Model A · fastUSE THISModel B · deepUSE THISAsk anything, or refine a response…

The system chooses

You write the prompt; it picks the mode and the model

There's no menu of routing modes or model names to get wrong. The system reads what you're asking and selects both the capability mode and the underlying model, so a simple question stays cheap and fast while a hard one gets the model it deserves.

  • System-chosen capability mode — no routing dropdown
  • Task-based model routing under the hood
  • The right model for the job, picked automatically
COMPARE · 2 MODELSYOUDraft a launch announcement for the new release.Model A · fastUSE THISModel B · deepUSE THISAsk anything, or refine a response…

Keys you control

Admin-managed provider keys, encrypted at rest

Admins enter your providers' API keys once, and they're encrypted at rest — never exposed to the browser and never shipped in the client. The workspace calls your providers server-side, so your keys stay yours and your conversations stay in your workspace.

  • Provider keys entered by admins, encrypted at rest
  • Keys never reach the browser or the client bundle
  • Conversations live in your workspace, ready to write up as docs
See docs & wiki
ENGINEERINGOverviewOnboardingSetupDeployRunbooksPostmortemsv12UPDATED BY MAYA · 2H AGOnpm run deploy --prod
“The part that stuck was not having to choose a model. People just ask, and the right one answers — the simple stuff stays fast, the hard stuff gets the heavy model. And our keys never leave the server.”
Rafael OtienoHead of Platform, Quanta Works

FAQ

Questions, answered

Do I pick which AI model to use?
No. You write the prompt and the system chooses both the capability mode and the underlying model based on the task — a simple question is routed to a fast, cheap model, a hard one to a stronger model.
Whose API keys does it use, and are they safe?
Your admins enter your providers' keys once. They're encrypted at rest, never shipped to the browser or the client bundle, and used only to call providers server-side.
Is my conversation data used to train models?
No. Your prompts and conversations are processed to serve your workspace only and are never used to train models. On a self-hosted deployment they stay on your infrastructure.
How is it different from using ChatGPT in a browser tab?
It lives in the same office as your meetings, chat, docs, and tracker — with multi-model chat, comparison, research, and image gen, automatic routing, and keys your admins control rather than personal logins.
Is the AI Workspace included in every plan?
The workspace is built into Remote Desk; plans differ in usage limits. See pricing for the per-plan AI usage included.

Ask, and the right model answers

The AI Workspace lives in the same office as your team — no model to pick, no keys to leak.

Start freeBook a demo