-
Start today

Chat Input Field

The chat input field at the bottom of the thread is where you type messages, attach files, choose agents, and control how the AI responds. This guide covers all input options in detail.

Key features

  • Text input — Multi-line input with Enter to send, Shift+Enter for new line
  • File attachments — Paperclip or drag-and-drop (documents embed into workspace; images attach to the message)
  • Agent selection — Choose specialized agents via button or @agentname
  • Model selector — Override the workspace model per thread
  • Web search toggle — Enable real-time web search for the thread
  • Slash commands — Presets and quick actions (e.g. /reset, /model)
  • Speech-to-text — Microphone for voice input
  • Pinned documents — Documents prioritized for this thread (shown above input when set)

Text input

  • Send: Press Enter.
  • New line: Press Shift + Enter or Ctrl/Cmd + Enter.
  • Focus on field: PressSpace.
  • Auto-resize: Input grows as you type (up to about 5 lines), then scrolls.
  • Placeholder: Shows "Send a message" when empty.

File attachments

How to attach

  • Paperclip — Click the attach icon, then select files
  • Drag and drop — Drag files onto the input area or chat

Attachment UI

  • Files appear as chips above the input
  • Shows status: uploading, success, or error
  • Click X on a chip to remove
  • Up to 10 files shown; extra files indicated as "+X more"

Agent selection

Agents are specialized assistants with distinct instructions and tools.

How to select

  1. Agent button — Click the agent icon in the toolbar; pick from the menu
  2. @mention — Type @ followed by the agent name; choose from suggestions
  3. Keyboard — Press Insert to open the agent menu

Agent pill

  1. Selected agent appears as a pill in the input
  2. Remove: Click X on the pill, or press Backspace at the start of the message
  3. Message is sent to that agent when you submit

Model selector

  1. Button — "Model:" in the toolbar opens the selector
  2. Options — Providers (OpenAI, Anthropic, etc.) and models within each
  3. System default — Use the workspace default model
  4. Scope — Selection is saved per thread

Web search toggle

  1. Button — Globe icon in the toolbar
  2. When on — AI can search the web for current information
  3. When off — Uses only workspace documents and model knowledge
  4. Use for — Recent events, real-time data, up-to-date facts

Speech-to-text

  1. Button — Microphone icon in the toolbar
  2. Shortcut — Ctrl/Cmd + M to toggle
  3. Flow — Speak; input shows transcript; message sends after a pause or when you stop
  4. Requirements — Microphone permission; browser with speech recognition support

Pinned documents

  1. Documents pinned to the thread appear above the input when configured
  2. Pinned docs are prioritized in AI responses
  3. Manage pins from Saved Items or the chat context menu

Keyboard shortcuts

| Key | Action | |-----|--------| | Enter | Send message | | Shift + Enter | New line | | Ctrl/Cmd + Enter | New line | | Escape | Unfocus input | | Backspace (start, agent selected) | Remove agent | | Insert | Open agent menu | | Ctrl/Cmd + M | Toggle speech-to-text |

Troubleshooting

  1. Input disabled — Wait for file uploads to finish, or check for errors
  2. Agent menu empty — Ensure agents exist and are assigned to the workspace
  3. Model not saving — Preference is per thread; re-select after switching threads
  4. Attachments fail — Check file type, size, and vision model for images
  5. Voice not working — Grant mic permission; use a supported browser

For more information, see:

  1. Chat Thread Page Guide — Overview of the chat page
  2. Chat Top Bar Guide — Thread actions and navigation
  3. Workspace Document Upload — How documents reach the workspace