CloudProvider connects your app to the Hiyve Cloud AI service. Provide a generateCloudToken callback that fetches a cloud token from your backend, and the SDK handles authentication and automatic token refresh:
Calendar component with switchable week and month views, event rendering, and navigation
CalendarWeekView
Week-grid calendar view with hourly slots and drag-to-create support
CalendarMonthView
Month-grid calendar view with multi-day event spanning
SessionList
Past meeting session history list with status, duration, and recording links
SessionCard
Single past session card with duration, participant count, and outcome
Mood Analysis
Component
Description
MoodAnalysisProvider
Provider that enables real-time facial emotion and engagement detection on participant video streams. Wrap your video components with this provider and use useMoodAnalysis() to access per-participant mood states. Configurable via analyzerType, detectionInterval, and smoothing props.
Sentiment Dashboard
Component
Description
SentimentDashboard
Real-time and historical sentiment overview
ParticipantMoodCard
Individual participant mood display
RoomMetricsOverview
Aggregated room engagement metrics
EmotionHeatmap
Emotion distribution heatmap over time
EngagementHeatmap
Engagement level heatmap over time
MoodSessionViewer
View saved mood session data
Intelligence Settings
Component
Description
IntelligenceSettings
Configuration panel for AI intelligence features
Intelligence Hub
Component
Description
IntelligenceHub
Tabbed container combining AI assistant, search, scorecard, and other intelligence panels
Search
Component
Description
SearchPanel
Semantic search across meeting transcriptions and documents
Scorecard
Component
Description
ScorecardPanel
Displays analysis scorecards (sales, interview, consultation, coaching, or custom)
Coaching
Component
Description
CoachingOverlay
Real-time coaching hints overlay during meetings
TalkRatioBar
Visual bar showing talk/listen ratio
TopicTracker
Displays current topic and topic shift history
Conversation Manager
Component
Description
ConversationList
Displays and manages saved AI conversation sessions
Agent Assistant
Components for building agent-style AI assistants with structured results, inline action forms, reasoning visualization, and network graph rendering.
Component
Description
AgentInput
Text input with send button for agent queries
AgentSidebar
Conversation history sidebar with create, rename, and delete
AgentConversationView
Scrollable message list with auto-scroll and structured results
AgentMessage
Single message display (user query or structured assistant result)
AgentResults
Renders structured assistant results including text, tables, lists, action buttons, forms, reasoning steps, and network graphs
ActionForm
Inline form for collecting missing parameters before executing an action
ReasoningSteps
Collapsible timeline showing the agent's reasoning process (tool calls, thinking steps, errors)
NetworkGraph
Force-directed graph visualization for relationship data
AgentInput Props
Prop
Type
Default
Description
onSubmit
(query: string) => void
required
Callback when a query is submitted
loading
boolean
false
Disables input and shows loading state
disabled
boolean
false
Disables the input
labels
Partial<AgentAssistantLabels>
-
Custom labels
colors
Partial<AgentAssistantColors>
-
Custom colors
styles
Partial<AgentAssistantStyles>
-
Custom styles
sx
SxProps<Theme>
-
MUI sx prop
AgentInput exposes an imperative handle via ref with setValue(text) and focus() methods.
A configurable card-grid dashboard designed for rooms without active video. The dashboard provides AI assistant, semantic search, alerts, room statistics, and settings cards in a responsive grid layout. Cards can be shown, hidden, reordered, or replaced with custom cards.
Component
Description
RoomDashboard
Orchestrator that arranges cards in a responsive grid
RoomStatsDashboard
Standalone full-page stats view with stat cards, chart slot, and recent files
DashboardCard
Reusable card wrapper with header, icon, badge, and expand/collapse support
AIAssistantCard
AI assistant card operating in stateless mode (no live meeting context)
SearchCard
Semantic search card for querying historical transcriptions and documents
Callback that fetches a cloud token from your backend. Called lazily on the first AI request and again automatically when the token expires. The SDK passes { userId } so your callback can forward it to your server. Can return a plain token string or an object with cloudToken and optional environment for automatic environment detection. Recommended approach.
cloudToken
string
-
A pre-fetched cloud token (skips generateCloudToken entirely). Useful when you manage token lifecycle externally.
userId
string
auto-resolved
User identifier (typically email) passed to generateCloudToken. Automatically resolved from IdentityProvider or HiyveProvider when available. Pass explicitly if you need cloud tokens before joining a room.
environment
'production' | 'development'
'production'
Target environment. When omitted and generateCloudToken returns { environment }, the environment is auto-detected from the token response.
baseUrl
string
-
Custom API base URL. Overrides environment if provided.
timeout
number
30000
Request timeout in milliseconds.
Key Types
Agent Assistant Types
Type
Description
AgentAssistantMessage
A message in an agent conversation (user query or structured assistant result)
AgentConversation
A conversation session containing messages and metadata
AgentReasoningStep
A step in the agent's reasoning process (thinking, tool call, tool result, error)
AgentReasoningStats
Statistics about the reasoning process (iterations, tool calls, duration)
GraphNode
A node in the network graph (id, label, type, color, properties)
GraphEdge
An edge in the network graph (source, target, type, weight)
GraphData
Graph data containing nodes and edges
ActionFormField
Field definition for an action form (name, type, label, required, options)
StoredRoom
A room entry for autocomplete in action forms
AgentAssistantLabels
Labels configuration for agent assistant components
AgentAssistantIcons
Icons configuration for agent assistant components
AgentAssistantColors
Color configuration for agent assistant components
AgentAssistantStyles
Style configuration for agent assistant components
AgentInputProps
Props for AgentInput
AgentInputRef
Imperative handle for AgentInput (setValue, focus)
Utility functions are prefixed by feature area to avoid naming conflicts when importing multiple features. For example, summaryFormatDuration and schedulingFormatDuration both format durations but for different contexts. Similarly, summaryGetPriorityColor and alertsGetPriorityColor return priority colors using each feature's color scheme. The prefix tells you which feature area the utility belongs to: summary*, alerts*, mood*, sentiment*, scheduling*, agent*, calendar*.
Type Aliases
Several common type names (e.g. ActionItem, MeetingAlert, Meeting, EmotionType) appear in multiple feature areas. To avoid collisions, import them using the prefixed aliases below:
Import Name
Description
CloudActionItem
Action items returned by cloud AI queries
IntelligenceActionItem
Action items displayed in the intelligence panel
SummaryActionItem
Action items displayed in meeting summaries
CloudMeetingAlert
Alert objects from cloud API responses
AlertsMeetingAlert
Alert objects used by alerts UI components
CloudMeeting
Meeting objects from cloud API responses
SchedulingMeeting
Meeting objects used by scheduling UI components
MoodEmotionType
Emotion values used by MoodAnalysisProvider
SentimentEmotionType
Emotion values used by SentimentDashboard
Utilities
Re-exported from @hiyve/cloud for convenience. These are available without importing @hiyve/cloud directly.
Cloud Client
Export
Description
CloudClient
Framework-agnostic client class for the Hiyve Cloud AI service
Validation
Export
Description
validateLength(text, max)
Validate that a string does not exceed a maximum length
validateRoomName(name)
Validate a room name format and length
validateTranscriptBatch(segments)
Validate a batch of transcript segments
MAX_QUERY_LENGTH
Maximum allowed query length
MAX_TRANSCRIPT_LENGTH
Maximum allowed transcript length
MAX_ROOM_NAME_LENGTH
Maximum allowed room name length
MAX_TRANSCRIPT_BATCH_SIZE
Maximum number of transcript segments per batch
Mood Bridge
Export
Description
emotionToSentiment(emotion)
Convert a mood emotion value to a sentiment entry
moodStateToSentimentEntry(moodState)
Convert a full mood state object to a sentiment entry
Cloud Defaults
Export
Description
DEFAULT_BASE_URL
Default API base URL
DEFAULT_TIMEOUT
Default request timeout in milliseconds
DEFAULT_MODEL
Default AI model identifier
DEFAULT_SUMMARY_MAX_LENGTH
Default maximum summary length
DEFAULT_ENHANCE_STYLE
Default text enhancement style
DEFAULT_ALERT_SETTINGS
Default alert configuration
DEFAULT_ENVIRONMENT
Default cloud environment
ENVIRONMENT_URLS
Map of environment names to API URLs
Formatting Utilities
Utilities exported from individual component modules:
Export
Description
formatAgentTimestamp(date)
Format a timestamp for display in agent messages
formatConversationDate(date)
Format a date for the conversation sidebar (today, yesterday, or date)
agentNormalizeRoomName(name)
Normalize a room name for display (removes random suffixes)
formatToolName(name)
Format a tool name for display (e.g., create_schedule to Create Schedule)
generateMessageId()
Generate a unique message identifier
formatMessageTime(date)
Format a message timestamp
groupMessagesByDate(messages)
Group messages by date for sectioned display
parseMarkdown(text)
Parse markdown text for rendering
summaryToMarkdown(data)
Convert meeting summary data to a markdown string
parseSummaryText(text)
Parse raw summary text into structured data
Heatmap Utilities
Functions for processing mood data into heatmap visualizations.
@hiyve/react-intelligence - AI intelligence components for Hiyve SDK
Remarks
This package consolidates all intelligence and AI-related components:
Example