EOne Documentation

website·Indexed 32 days ago

https://docs.eone.vn

An intelligent platform for building AI agents to revolutionize customer interactions. EOne provides tools for creating 24/7 customer service and sales agents, featuring a Python and Node.js SDK, Model Context Protocol (MCP) integration for custom tool calling, and a unified channel dashboard for managing communication channels, telephony, and LLM-powered orchestration.

Tokens
25.4K
Snippets
66
Records
198
Agent score
49%

What's inside EOne

  1. Overview of EONE platform capabilities for developers

    EONE is an API-first, intelligent customer interaction platform designed as a programmable layer for company conversations. Developers can use EONE to:

    • Sync CRM/ERP Data: Push call logs, transcripts, and AI-extracted data into backend systems.
    • Extend AI Agents: Use the Model Context Protocol (MCP) to provide AI Agents with tools for real-time actions like checking order statuses or booking appointments.
    • Automate Data Entry: Automatically extract structured information from conversations and send it to a specified endpoint.
    • Trigger External Workflows: Start workflows in external systems based on conversation events (e.g., detecting negative sentiment).
  2. Overview of the EOne omni-channel platform

    EOne is an omni-channel contact center platform with an AI core that unifies customer conversations from Voice (Phone Calls), Zalo, and SMS into a single intelligent inbox. It is designed to provide a 360-degree view of customers, allowing AI Agents and Human Agents to collaborate seamlessly to move from reactive problem-solving to proactive engagement.
  3. Retrieve conversation details by ID

    Use the Get conversation detail endpoint to fetch detailed information about a specific conversation. The request requires a conversation ID passed as a path parameter.
    # Example request structure
    GET /api/get-conversation-detail/{id}
  4. Understand the role of Knowledge Bases in EONE

    A Knowledge Base in EONE allows an agent to access large volumes of external data—such as product documentation, FAQs, website content, and articles—without bloating the prompt. This mechanism enables agents to answer specific questions accurately and consistently while reducing hallucinations. The system works through three main stages:

    1. Data Source Integration: Connecting EONE to information sources via document uploads, website URLs, or database providers.
    2. Indexing: Processing data into efficient search formats, typically using vector embeddings.
    3. Retrieval: The LLM automatically queries the index when it identifies a user question that requires external information.
  5. Understand the EONE Telephony Component

    The Telephony component serves as the gateway between EONE agents and the public telephone network (PSTN) or voice communication systems (such as SIP). It is responsible for managing phone numbers, inbound call routing, outbound call initiation, and call control signaling. Key capabilities include:

    • Provider Integration: Connecting accounts from providers like CCALL, Viettel, VNPT, Mobifone, iTel, GTEL, Twilio, or using EONE native options.
    • Number Management: Purchasing, configuring, and assigning phone numbers to specific agents or workflows.
    • SIP Trunking: Establishing direct SIP connections for advanced integrations.
    • Call Routing: Defining rules for directing incoming calls.
    • Outbound Call Settings: Configuring caller ID and other parameters for outgoing calls.
  6. Understand and implement EONE Webhooks for real-time event notifications

    Webhooks in EONE are automated POST requests sent to a specified external URL when specific events occur within the platform. They enable asynchronous reactions to platform events, such as logging call data to a database, triggering follow-up actions, or updating external dashboards.

    To implement webhooks, you must configure:

    1. Event Subscriptions: Select the specific events to monitor. Common events include:
      • call:started: Triggered when a call begins.
      • call:ended: Triggered when a call finishes.
      • transcript:ready: Triggered when a call transcript is available.
      • analysis:ready: Triggered when call analysis is complete.
      • function:called: Triggered when a platform function is invoked.
    2. Endpoint URL: A secure URL in your application capable of receiving POST requests.
    3. Security: Implement signature verification to validate that incoming requests originate from EONE.
  7. Monitor KPIs using the Main Performance Dashboard

    The Main Performance Dashboard provides a high-level view of operational health through several key widgets:

    • Interaction Volume: Total conversations categorized by channel (Voice, Zalo, SMS).
    • Automation Rate: Percentage of conversations fully resolved by AI Agents without human intervention.
    • Average Handle Time (AHT): Average time taken by a human agent to resolve a conversation.
    • First Contact Resolution (FCR): Percentage of issues solved in the first interaction.
    • Overall Customer Sentiment: Aggregated emotional tone score of interactions over a selected period.
  8. Understand the role of Agents in EONE

    An Agent in EONE is a central entity representing a specific conversational AI persona designed for a particular task (e.g., Appointment Booker, Tier 1 Support, or Lead Qualifier). The Agent configuration acts as the orchestrator that ties together the prompt, AI models, voices, tools, knowledge sources, and conversation flow parameters.
  9. Navigate the EOne 'My Work' Dashboard

    The 'My Work' dashboard is the default landing page upon logging into EOne. It serves as a daily briefing and priority overview, containing three primary components:

    • My Open Conversations: A list of assigned conversations requiring immediate action.
    • My Tasks Due Today: A checklist of follow-up tasks due for completion.
    • My Performance Snapshot: A summary of the agent's key performance indicators (KPIs).
  10. Understand EONE's High-Level Architecture

    EONE acts as a central hub that unifies communication channels and enhances them with AI. Developers interact with the platform via a REST API, Webhooks, and the Model Context Protocol (MCP). The system is designed as an orchestration engine that manages the lifecycle of real-time interactions (voice and chat) by routing data between communication interfaces, AI processing pipelines, and external integrations.
  11. Analyze conversations with Conversation Intelligence

    EOne automatically analyzes every conversation to provide deeper insights beyond basic metrics:

    • Sentiment Analysis: Interactions are tagged as Positive, Negative, or Neutral to identify unhappy customers and root causes.
    • Topic & Intent Detection: The AI identifies the primary subject and purpose of conversations (e.g., 'Questions about delivery status' or 'Intent to cancel') to spot emerging trends.
  12. Understand the role of Synthesizers in EONE

    The Synthesizer, also known as the Text-to-Speech (TTS) engine, is the component responsible for converting text responses generated by a Large Language Model (LLM) into audible speech. The choice of synthesizer affects the agent's professionalism, engagement, and the overall conversational flow, with latency being a critical factor for natural interaction.