Resume and Steer AI Chats from Mobile
Mobile project chat connects to agent sessions on your paired desktop. Use the correct project and provider to resume a conversation, add context, or follow a running task without starting over.
Open the project and choose the agent
Open the project’s AI chat and select an available provider, such as Claude Code, Codex, Gemini, or Grok. Choose the model and any reasoning options the picker offers. Available options come from the desktop and vary by model; configure the provider there if it is unavailable.
Find a previous conversation
Open Past chats using the history control. The list is scoped to the current project and provider and starts at Today. Choose Last 7 days, Last 30 days, or Custom to pick a day. Dates are based on last activity, so a chat started earlier may appear on the day you last used it. Use Load more for additional results.
Resume with the right context
Select a chat and review its transcript before sending your next instruction. Use Library context to choose blocks or projects, and use the plus menu to attach an image when supported. Wait for attachments to finish uploading before sending. Describe what the agent should do with the attached material.
Send a follow-up while the agent works
Read the composer’s current action. Send to active chat steers a supported running session; Add to queue saves an instruction for the queued workflow. These behaviors depend on the provider and desktop capability. Check that the follow-up appears in the conversation or queue, and update the desktop if live steering is reported as unavailable.
Reconnect and verify completion
After switching screens or recovering a connection, reopen the same project and chat to review its current status, tool activity, and final reply. A temporary transport interruption does not prove that desktop work stopped. If Past chats is empty, check the selected provider and date filter; use Retry for a loading error before creating a duplicate task.