Package-level declarations
Types
Parameters for ChatAiComponentFactory.AIConversationAssistantMessage.
Parameters for ChatAiComponentFactory.AIConversationAttachment.
The palette of AIConversation's own views.
Defaults of AIConversation.
The icons of AIConversation's own views.
Parameters for ChatAiComponentFactory.AIConversationReplyActions.
Parameters for ChatAiComponentFactory.AIConversationUserMessage.
The palette of AIMessageAttachmentItem.
Defaults of AIMessageAttachmentItem.
The icons of AIMessageAttachmentItem, one for each type of file. A file's type comes from its media type, else from the extension of its name or URI.
The palette of AIReplyActionsRow.
Defaults of AIReplyActionsRow.
Parameters for ChatAiComponentFactory.AISuggestionsChip.
The palette of AIToolApprovalCard.
Defaults of AIToolApprovalCard.
Where the person's answer to a question is.
The person answering tool calls' questions and sign-ins on this device, and how their answer reaches your backend, which holds each call until it gets one.
Parameters for ChatAiComponentFactory.AIToolAuthorizationContent.
How the person signs in on this device when a tool call waits for their account at a provider, and how their "not now" reaches your backend, which holds the call until one or the other.
The palette of AIToolCall.
Defaults of AIToolCall.
Parameters for ChatAiComponentFactory.AITypingIndicatorIndicator.
Parameters for ChatAiComponentFactory.AITypingIndicatorLabel.
The palette of AIUserMessage.
Defaults of AIUserMessage.
What a voice agent is doing, which a voice visualization shows as motion.
How AIVoiceGlow is drawn. Either way it moves the same; only the look and the cost differ.
Defaults for the voice visualizations.
Factory that creates the components rendered inside the AI chat UI.
Parameters for ChatAiComponentFactory.ComposerInputContent.
Parameters for ChatAiComponentFactory.ComposerInputTrailingContent.
Parameters for ChatAiComponentFactory.ComposerLeadingContent.
Parameters for ChatAiComponentFactory.ComposerTrailingContent.
Data class representing a message composed by the user.
Parameters for ChatAiComponentFactory.SpeechToTextButtonIdleContent.
Parameters for ChatAiComponentFactory.SpeechToTextButtonRecordingContent.
State holder for SpeechToTextButton that tracks the recording status.
The palette of StreamingReasoning.
Defaults of StreamingReasoning.
Properties
The composition local that provides the current ChatAiComponentFactory.
Functions
A reply from the agent: the steps it took, then its answer, written out as it arrives, and then its attachments, across the whole width.
Shows AIMessages with the views of LocalChatAiComponentFactory: each message the person sent with ChatAiComponentFactory.AIConversationUserMessage, each reply with ChatAiComponentFactory.AIConversationAssistantMessage, and under a finished reply, once its answer is written out, ChatAiComponentFactory.AIConversationReplyActions.
The messages of a conversation with an AI agent, oldest first, shown your own way.
One of a message's attachments: an image as a square thumbnail, and any other kind as a file, with its name, type and size, that opens when tapped.
One step of a reply: a round of reasoning with its preview, a tool call (with its question or its sign-in, when it waits for the approver), or a neutral placeholder for a step this SDK doesn't know. The sign-in shows through ChatAiComponentFactory.AIToolAuthorizationContent.
The steps an AI agent took while replying, in order, each shown by content.
The steps an AI agent took while replying, in order. Show it before the reply's text.
The buttons under a finished reply: copy and share, then regenerate and rate when the app handles them.
A horizontally scrolling row of suggestion chips, usually placed above ChatComposer on a new chat.
A reply's thinking before any of it arrives: "Thinking…", counting the seconds.
A tool call's question: what it asks, the agent's reason and what allowing it shares, and buttons to allow or decline it.
Asks the person a tool call waits for whether it may run: the question its ai_tool_call step carries (status awaiting_approval, with approval). It shows only to that person, on the device the call names, and only while the call waits, so it can sit under every tool call:
A tool call's sign-in: which provider it needs and what signing in means, with buttons to connect or to say not now.
Asks the person a tool call waits for to sign in at the provider it needs: the sign-in its ai_tool_call step carries (status awaiting_authorization, with authorization). Connect hands the sign-in page to the approver's AIToolAuthorizer, which opens it. Once the provider has answered, your backend updates the step and the call goes on. It shows only to that person, only while the call waits, and only with an AIToolAuthorizer, so it can sit under every tool call. Once the sign-in's AIToolAuthorization.expiresAt passes, it says so and stops offering Connect:
One tool call: what it is doing, where it runs, and how it went.
A typing indicator composable that displays an optional label and an animated indicator.
A message the person sent: its attachments, then its text in a bubble, at the end of the row. Pressing and holding the bubble offers to copy the text, and to edit it given onEdit.
A voice agent as a soft, airy globe: color masses orbit inside it and swell with speech, and the gaps between them let the background show through. Tapping the globe bounces it; drawn with the shader, it also ripples from the touch and under a hovering mouse or stylus.
A voice agent as a single string, flat at its ends, carrying a wave packet that erupts with speech while its colors flow along it. Tapping the string ripples it from the touch, and so does a hovering mouse or stylus.
Chat composer with attach, voice, and send buttons.
Chat composer with attach, voice, and send buttons, whose message the caller owns.
Builds a compound ChatAiComponentFactory by wrapping the current factory and provides it to content through LocalChatAiComponentFactory.
Creates and remembers a SpeechToTextButtonState instance.
A composable button that provides speech-to-text functionality.
Shows a reasoning step: its live text when the app has it, otherwise the step's preview, with its summary beside "Thought for 12s" once it is done.
A model's reasoning (its "thinking"), shown alongside its reply.
A composable that wraps content and provides streaming animation. Progressively reveals text content word-by-word and whitespace chunk-by-chunk.