The Workspace,
piece by piece.
Below is the Workspace as it actually appears, with each part numbered. Tap any number to jump to its dedicated section — sidebar, mode picker, model picker, composer, credit pill.
Sidebar.
The sidebar is the entry point to everything in AiFuse:
- New chatStarts a blank conversation. Mode and model default to whatever you last used.
- ProjectsLong-lived workspaces. Threads, files, and saved prompts grouped by what you're working on.
- Art StudioImage and video generation. Coming soon.
- Talk to AIRealtime voice agents. Coming soon.
- HistoryEvery chat you've had, searchable. Click the pen icon to rename a chat; double-click the trash icon to delete it.
- Theme toggleSwitches between light and dark.
- CollapseClick the « to shrink the rail to icons.
- User avatarClick to open a menu with Usage, Settings, Help, Support, and Log out.
Mode picker.
Pick a mode when you start a new chat. Four to choose from. The mode stays fixed for that conversation:
- AutoProvider and model is automatically selected based on what you ask. The default for new chats until you've used another mode.
- SingleSelect your desired provider and model — for when you know what you want.
- DuoTwo providers, one prompt. Answers stream side by side so you can compare quality and price. Pick the one to carry forward as context for the next turn.
- FusionPick two providers and models. The first acts as Orchestrator, the second as Consultant. The Orchestrator uses input from the Consultant to deliver a more refined answer.
Provider / model / context picker.
Based on the selected mode, all or some of the following settings apply:
- TierOnly in Auto mode. Constrains which models the auto-router can pick. Lite favors cheapest, Balanced excludes premium flagships, Pro allows everything.
- Context
Controls how much prior conversation is sent with each new message. More context means the model remembers more, but each turn costs more to send. Four levels:
- None — current message only.
- Economy — last 10 messages.
- Balanced — last 25 messages. Default for new chats.
- Quality — last 50 messages.
Lives in the model picker modal (Single, Duo, Fusion) and Auto's composer chip. In Duo and Fusion, only the primary slot shows it — the secondary uses the same setting.
The provider and model picker opens from the composer chip. Browse two ways:
- By provider — models grouped under their provider.
- By task — models grouped by what they're good at, like coding, image generation, or voice.
Available in Single, Duo (primary slot), and Fusion (orchestrator slot). Auto picks the model for you.
Composer.
The composer sits at the bottom of every chat. Type your message, then send with the gold button on the right, or ⏎.
- AttachDrop files into the composer: PDFs, images, Word, Excel, code, transcripts. Support varies by model.
- Prompt BuilderA 5-step modal for structured prompts — area, role, task, style, preview. Use it when you want a careful prompt.
- IncognitoToggle to send a turn that isn't saved anywhere — no local history, no record on our servers. The message is lost on reload or when starting a new chat. The provider still sees the message.
- ⏎ to sendOr click the gold button. Shift+⏎ for a newline.
