What are OCP® Agents

Overview

An Agent in the Omilia context is an autonomous or semi-autonomous component designed to perform tasks within a contact center ecosystem. Agents integrate perception, reasoning, and action to handle customer interactions, automate workflows, and provide intelligent routing and responses.

Core Capabilities

Agentic Intelligent Routing (Concierge Agent)

The Concierge Agent provides a new paradigm for contact center routing by using real-time reasoning and contextual awareness instead of static queue logic and intent matching.

• Zero-Shot Routing: Dynamically routes calls or chats to the optimal queue or agent without pre-programmed mappings (no intents)

• Dynamic Queue Awareness: Continuously monitors queue load, agent skill, and conversation intent to make intelligent routing decisions. It is fully integrated with the Contact Centre to ensure real-time adaptability.

• End-to-End Orchestration: Acts as a control layer across all channels and agents. Defines the agents, flows and context that should be through the entire conversation.

• Seamless PhoneBook Integration: Supports routing based on PhoneBook entries to queues or extensions.

• Automatic Disambiguations: Automatically detects missing or important information and prompts the callers before queue transfer.

This ensures minimal transfer rates, higher first-call resolution, and optimized operational efficiency. It enables a zero-day go live as it requires no training.

Task Agents

Task agents define all of Omilia’s Agents (regardless of their level of autonomy) designed to perform tasks. They perform multi-step workflows, integrating perception, reasoning, and action within the contact center ecosystem.

Agents can have three different levels of autonomy:

• Low Autonomy

• Medium Autonomy (Fallback)

• High Autonomy

Low autonomy Agents

Fully powered by miniApps, it offers fully controlled and explainable agents to handle tasks that can be put together in a well determined flow. Applications designed to handle specific, well-defined tasks within a rigid and predictable workflow.

Medium autonomy agents

Omilia’s Medium Autonomy Agents (a.k.a Agentic Fallback) integrate seamlessly with existing deterministic (Low Autonomy) workflows (such as those built in Orchestrator and miniApps™) to enhance understanding and improve containment rates. There are two levels of fallback support:

1. Asynchronous context update - The agent is updating the context of the conversation based on a wider understanding of the conversation.

2. Full-handover - The Task Agents attempts to complete the task.

Medium autonomy agents are characterized by the following:

• Adaptive Understanding: When predefined flows fail to capture a user’s intent or fail to fulfill the requested task, the Agentic AI dynamically interprets the request using contextual reasoning, enterprise knowledge, and a broader understanding of the context.

• Intelligent Recovery: Rather than failing silently, the system activates a fallback agent that can clarify, infer, or autonomously complete the task based on business rules and guardrails.

• Safe Escalation: If ambiguity remains, the fallback agent escalates intelligently to human agents or specialized systems with complete conversation context (through Intelligent Routing).

• Determined intelligence: The boundaries and context of the AI agent are well contained within the task at hand, not allowing unbounded interactions with no visibility.

This approach bridges the gap between static workflows and autonomous AI, ensuring resilience and uninterrupted service continuity.

High Autonomy Task Agents

High Autonomy Task Agents are specialized autonomous agents designed to execute multi-step workflows, integrating perception, reasoning, and action within the contact center ecosystem.

High autonomy agents are characterized by the following:

• Dynamic Planning & Execution: These agents autonomously plan and execute tasks in real-time without relying on pre-built reactive workflows.

• Integration-Ready: They interact with CRMs, billing systems, and ticketing tools to automate complex, cross-system processes end-to-end. 

FAQ Agents

Lightweight agentic components specialized in retrieving, synthesizing, and delivering accurate answers based on existing documents or websites. They extend the Omilia Agentic Framework’s reasoning layer to provide instant, contextually grounded responses.

• Retrieval-Augmented Understanding: The FAQ Agent leverages customer’s documentation, and Omilia’s RAG (Retrieval-Augmented Generation) layer to extract accurate and explainable responses from structured and unstructured sources.

• Adaptive Answering: Rather than serving static FAQs, the agent dynamically composes personalized responses, taking into account conversation context, customer profile, and intent confidence.

• Optimized for voice:
  Operates across voice and digital channels, but with latency optimized for voice.