​
AI Agents Overview

AI Agents are the functional core of the Plura platform. They are the assistants that handle live conversations, respond to user input, and guide interactions across different workflows. While workflows define what the assistant should do, agents define how it behaves—what it sounds like, how it responds, and how it connects to users.

An AI Agent brings together various configurations that control its behavior, personality, voice, and the call routing logic. This page provides a high-level overview of AI Agents and the various features that can be managed through the agent settings.

​
What Are AI Agents?

AI Agents in Plura represent deployable assistants that can interact with users over phone or chat. Each agent has its own configuration that determines how it behaves during conversations, which workflow it follows, and how it manages technical settings such as voice options or call routing.

They are best understood as the “personality layer” on top of your workflow logic. Multiple agents can share the same underlying workflow, but differ in how they speak, what phone number they use, or what analytic evaluations they perform.

​
Agent Dashboard

The AI Agent dashboard provides a centralized overview of all existing agents in your organization. Here, you can:

  • Create and configure new AI Agents
  • Assign a workflow to each agent (previously labeled “journey”)
  • Assign inbound and outbound phone numbers
  • View current status and last modified dates
  • Choose from a wide range of synthetic voices

This dashboard offers a quick glance at key agent properties like active status, voice selection, and phone assignment.

​
Key Configuration Areas

Each AI Agent has multiple configuration tabs that allow for fine-tuning of its functionality. While we won’t go into deep detail in this overview, here’s what each section is responsible for:

​
Configuration

  • Assign inbound and outbound workflows
  • Set phone numbers for inbound and outbound calling
  • Add a short description to help identify the purpose of the agent

​
Voice Settings

  • Choose the AI voice provider and voice style
  • Configure audio behavior, like background noise simulation
  • Filter voices by language, gender, and use case

Agents can be tailored to match specific demographics or brand tones by selecting different voices and tuning vocal behavior.

​
Analysis

  • Set up automated prompts to evaluate the call’s outcome, summarize interactions, or extract structured data
  • Helps with post-call insights and quality control

This feature allows you to programmatically review and analyze every AI-handled call using system-defined or custom prompts.

​
Advanced Settings

  • Configure privacy-related settings like HIPAA compliance and audio recordings
  • Define behavior like silence timeout or response delay

This section controls how strictly the agent handles data, whether to record conversations, and how quickly it responds or disconnects during user silence.

​
Deploy

  • Add optional Webhook URL to forward call data
  • Set voicemail fallback message
  • Access API endpoint structure for manually triggering agent calls via HTTP

This enables full integration with external systems, allowing data to flow in and out of Plura seamlessly.

​
Calls

  • View the agent’s historical call logs
  • Filter by call direction (inbound or outbound), duration, and more

A simple but essential tool to monitor ongoing agent performance.

​
Summary

AI Agents are highly customizable and modular, acting as the interface between your workflow logic and real-world user interaction. With features ranging from voice control to privacy settings and analytics, they allow you to deploy scalable, intelligent voice and chat assistants with precision.

The Agent Overview page gives you full visibility and access to manage everything that makes an assistant unique—from the voice it uses to the logic it follows.

In the following pages, we’ll dive deeper into each individual tab to show how to configure and fine-tune your AI Agents for different use cases.