Voicebot by UnifyApps enables sophisticated voice interaction capabilities within automation workflows.
Overview
Voicebot by UnifyApps enables sophisticated voice interaction capabilities within automation workflows. This integration allows creation, management, and processing of voice-based communications through AI-powered voice agents, providing real-time voice processing, session management, and intelligent response handling for customer service automation, voice-activated workflows, and interactive voice response systems.


Use Cases
Automated Customer Support — Customer service teams implement Voicebot to handle initial inquiries. When customers call, the voicebot starts a session, listens to requests, processes audio content, and provides intelligent responses. Complex queries transfer to human agents with full conversation context, reducing wait times and improving satisfaction.
Voice-Activated Workflow Triggers — Logistics companies use voice commands to trigger warehouse operations. Workers speak into devices to initiate inventory checks, shipment processing, or status updates. The voicebot processes voice commands, converts them to actionable data, and triggers appropriate workflow sequences.
Interactive Voice Surveys — Market research firms automate survey collection through voice interactions. The voicebot initiates sessions with respondents, asks survey questions, processes spoken responses, and records answers in structured formats.
Typical Workflow Pattern
The node supports two flows:
Web sessions — Start Voicebot Session, then Start Listening to Voicebot Session, then Respond to Voicebot Session. These actions are linked by the Session Id and Room Name.
Call requests — The On New Request trigger starts the automation for each new request, and Send Response delivers the reply. Send Response uses the Channel Name and Call Sid from the trigger.
To run an AI voice agent in a live audio room, use Start Voice Agent Session or Start Voice Team Agent Session.
Send Response
Enables automation to provide voice responses back to users during active voice sessions, controlling response delivery and call flow.
Key Input Fields:
Channel Name (required) — The channel to send the voicebot response to. Map it from the On New Request trigger.
Call Sid (required) — The call to respond to. Map it from the On New Request trigger.
Voicebot Response — The response to send.
Interrupt Prompt (Boolean, default: true) — Stops the current voicebot prompt in progress, if any. Set it to false to let the current audio finish.
End Call (Boolean, default: false) — When true, the call is disconnected after this response.
Output: sent (Boolean).


Respond to Voicebot Session
Processes incoming voice interactions and prepares responses within an active voice session, serving as the core processing engine for handling user voice input.
Key Input Fields: Event Id (String); Audio Content (String); Room Name (String); Session Id (String); Received Time and Sent Time (Integer); Start Event (Boolean); Agent State (String); Format (String). None of the fields is required.
Output: success (Boolean).


Start Voicebot Session
Initializes new voice interaction sessions, establishing connection between users and voice-enabled automation workflows.
Key Input Fields: Session Id; User Id; Workflow Id; Real Time (Boolean, default: true).
Output: room_name, room_id, token and server_url — the connection details for the live audio room.
Start Listening to Voicebot Session
Activates audio input monitoring for established voice sessions, enabling the voicebot to actively listen for user voice input.
Key Input Fields: Session Id (required); Workflow Id (required); Room Name; Track Id (for multi-stream environments).
Functionality: Audio Stream Connection; Voice Activity Detection; Noise Filtering; Continuous Monitoring; Event Triggering.
Output: success (Boolean).
Start Voice Agent Session
Starts a new LiveKit voice agent — an AI that listens, understands and speaks back over a live audio session. You configure the speech-to-text (STT), large language model (LLM) and text-to-speech (TTS) providers when you start the session; they aren't selected automatically.
Other settings include the turn detection mode (STT, VAD (default), Realtime LLM or Manual), whether callers can interrupt the agent (on by default), noise cancellation (Krisp or Noisereduce, off by default), the language (default en), welcome message instructions, and whether the session closes when the caller disconnects (on by default).
Output: room_name, room_id, token and server_url — the connection details for the live audio room.
Start Voice Team Agent Session
Starts a new LiveKit voice team agent, where a team of agents handles the conversation. Like Start Voice Agent Session, you configure the STT, LLM and TTS providers, the language and the welcome message instructions, and you list the agents in the team.
Output: room_name, room_id, token and server_url.
On New Request (trigger)
Starts the automation each time a new voicebot request arrives. You can add trigger conditions so only matching events are processed.
Output: channelName, callSid, type, content and aiAgentId. Map Channel Name and Call Sid into Send Response to reply to the request.
Error Handling Strategy
All voicebot actions are configured with a "STOP" fallback mode. Errors in any voice action halt the workflow to prevent partial voice sessions or confused user experiences. Proper error handling ensures voice interactions remain coherent throughout the session lifecycle.
Note: Caching and Retry are disabled by default for all voicebot actions, as real-time voice processing requires fresh data and duplicate session creation must be prevented.
Notes
Keep the following in mind when using Voicebot by UnifyApps:
Keep the two flows separate: web-session actions use the Session Id and Room Name, while Send Response uses the Channel Name and Call Sid from the On New Request trigger.
Retry and caching are disabled for all voicebot actions by default; do not enable them — duplicate session operations produce unpredictable behavior.
Set End Call to true in Send Response only when the conversation is complete; leaving an active session open unnecessarily consumes resources.
Use the Interrupt Prompt field in Send Response to control audio playback; set it to false when you want the current audio to finish before delivering the next response.
Test voicebot automations end-to-end in a staging environment before deploying; partial session execution is difficult to diagnose from run logs alone.