Overview
Ask AI brings the power of artificial intelligence to your workplace, providing instant answers to questions, intelligent insights, and automated assistance. The AI assistant can help with policy questions, process guidance, data analysis, and general workplace inquiries.
Highlights
Capabilities
Conversational Interface
-
Natural-language chat from web and mobile
-
Persistent conversation history across devices
-
Multi-turn follow-up questions with context
-
Multi-lingual greetings and response translation
-
@mention a specific agent inline (e.g. @scheduling)
-
Streaming responses via WebSocket (ActionCable)
-
Collaborative shared conversation history One active thread per user
Agent Routing
-
Automatic intent classification via hybrid ML + LLM
-
Context-aware routing for short or ambiguous messages
-
Sticky agent routing for multi-turn domain sessions
-
Pre-classification from streaming job for low latency
-
Reporting agent fast-path (pattern + semantic match)
-
Plugin agent proxy support (third-party extensions)
-
Number of built-in domain agents: 60+
RAG & Knowledge
-
Answers grounded in company knowledge base articles
-
System help articles included in RAG context
-
Conversation fact extraction for long-term memory
-
Embedding-based context retrieval with cache
-
Token-budget management (128 k context window)
-
Conversation summarizer for long threads
-
Real-time data from connected apps (schedules, PTO)
Tool Calling & Actions
-
Agents can call tools to read live data
-
Write actions require explicit user confirmation
-
Human handoff โ escalate to HR / IT / service desk
-
Auto-create service desk ticket when AI cannot answer
-
Chain-of-thought / deep reasoning mode per agent
-
Agentic parameter collection (multi-step clarification)
-
Agents execute multi-step payroll / scheduling tasks without confirmation Confirmation always required for writes
Voice & Multimodal
-
Voice mode with speech-to-text input
-
Text-to-speech (TTS) playback of AI responses
-
LLM-based voice conversationalization (natural speech)
-
Configurable voice response style (concise / detailed)
-
Per-business voice mode enable / disable toggle
-
Real-time (OpenAI Realtime API) voice conversation
-
Video or image input in chat Text and voice only
Safety & Audit
-
Content moderation with prompt-injection detection
-
Business-scoped tenant isolation (data never crosses orgs)
-
All LLM calls logged to AiApiLog with session ID
-
Sentinel rules โ custom alert triggers on AI metrics
-
Security system-prompt injected on every request
-
Admin toggle to disable Ask AI per business
Limits & Specs
-
Voice mode daily limit (non-dev): 30 minutes per user
-
Voice session limit: 15 minutes per session
-
LLM context window: 128,000 tokens
-
ML router confidence thresholds: High โฅ 0.3 ยท Medium โฅ 0.15 ยท Low < 0.15
-
Conversation history retention: Single active thread per user (admin-clearable)
-
Pricing: Included with MangoApps Workforce
Use cases
FAQ
Responses are grounded in your organization's knowledge base, uploaded documents, company policies, and live data from connected apps.
Specialized agents can take actions like generating reports, looking up schedules, and checking leave balances. Actions that modify data require confirmation.
Conversations are scoped to your business tenant. Data is never shared across organizations and all queries are logged for audit purposes.
The assistant indicates when it lacks sufficient information and can suggest contacting HR, IT, or the relevant department directly.