How to drive Toad in Studio.

One conversation with the brain, with the work docked around it.

• Role daily surface
• Brain almost any model
• Approvals inline in chat
• Proof docked Workbench panels
• Sessions resume per workspace

1. Type the work — describe what you want in plain language.
2. Let Toad route it — background agents by default; steer with your wording.
3. Approve inline — plans, memory, and code land in the conversation.
4. Open proof — dock a Workbench panel beside the chat.
5. Resume anywhere — each workspace keeps its own session.

studio — chat and live agents

Type the work in the composer

• Say it plainly — a one-line fix, a whole feature, a codebase question.
• Toad plans it, then answers directly or delegates to agents.
• Follow-ups stay in the same thread; context carries over.
• The brain runs on almost any model — CLI subscription, API key, or local — but not the Gemini/Antigravity CLI (agents-only). → Accounts

Choose where the work runs

Toad decides the default routing; steer it with the words you use.

Approve plans, memory, and code inline

• Plans, memory, code, and doc changes surface right in the conversation.
• Approve or redirect in context — no separate queue to hunt through.
• Memory is never written silently: evidence-linked, saved only when you approve.

Open Workbench panels for proof

• Open Preview for a rendered page, Files and Git for source, Terminal for a focused check.
• Panels dock beside the chat — you and Toad inspect the same evidence.
• GitHub actions run via the gh CLI, using your existing login. → Workbench

Resume across workspaces

• Each switch reloads that workspace’s session state and brain memory.
• One workspace is active at a time; the rest keep their history.

Talk with voice when it is faster

• Voice and live-voice controls sit beside the composer, including the xAI realtime lane.
• Toad transcribes what you said and drives the same chat loop.