Genesys Chatbots

Use chatbots to automate customer conversations and seamlessly hand over to a chat agent when needed.

What's the challenge?

Many customer service, sales or support conversations with customers are repetitive — frustrating both to customers and to employees. If you could insert better automation, many conversations may well be taken care of in the entry process, saving time while also increasing customer satisfaction.

What's the solution?

Blended AI chatbots automate natural language conversations, even across channels. Genesys blended chatbots look up customer information and activity to answer questions. They can hand over conversations with context to an agent when needed, or even offer a callback during or after hours.

Use case overview

Story and business context

The proliferation of digital channels leads to higher customer expectations and an increased number of interactions that companies deal with when servicing customers. Coupled with increased usage of Artificial Intelligence (AI) for business applications, this change results in organizations implementing chatbots that can interact with customers to automate tasks and assist their queries on digital channels such as web, mobile, social, SMS, and messaging apps. Chatbots can alleviate strain on contact center employees while improving the customer experience and controlling costs. Chatbots are always on and available, and can hand over to a live agent at any time where needed. While chatbots can also be used by employees and for business optimization purposes, the remainder of this document refers to omnichannel bots in the context of customer engagement. The primary benefits of chatbots are to increase self-service success, deflect interactions from the contact center, and improve the customer experience.

Genesys chatbots unify and orchestrates self-service experiences using both native and third-party bots – powering exceptional customer and employee experiences. Genesys supports a “design once, deploy anywhere” concept for bots to enable organizations to provide a seamless customer experience across voice and digital channels. This use case focuses on deploying a bot on web chat, mobile chat, Facebook Messenger, Twitter Direct Message, Line Messaging, WhatsApp, or SMS.

Use case benefits

Benefit Explanation
Improved Containment Rate Increase self-service interactions to reduce agent-assisted interactions for repetitive or common requests.
Improved Customer Experience Reduce the time required to address the customer request, handle off-hour contacts, offer immediate options, and improve outcomes.
Improved First Contact Resolution Tailor the customer experience to the individual based on who they are, why they could be interacting, and the status of the contact center

Summary

Genesys Chatbots supports native platform Dialog Engine Bot Flows and third-party platforms such as Amazon, Google etc. As each chatbot and third party has their own specific capabilities, this use case covers broadly available capabilities, for the most of to date latest references available, visit the Resource Center.

The chatbot supports or orchestrates the following capabilities:

  • Personalization – to tailor the experience based on context from the current interaction or from previous interactions
  • Natural Language Understanding – to derive intents and entities
  • Simple bot orchestration enables customers to use the best bot for the job. For example Google Dialogflow has highest alphanumeric recognition rates
  • Genesys Cloud CX Architect makes it easy to integrate to new bot providers, switch between bot providers or to use multiple bot providers within a single interaction
  • A-B testing with Genesys Cloud CX Architect helps determine which bot is most effective for a particular business use case
  • Graceful escalation to a live Agent at the right time

Use case definition

Business flow

When a customer interacts through a supported Genesys digital channel, a chatbot starts. The chatbot first attempts to use context to anticipate why the customer may be engaging and in turn provides personalized messages to resolve the query. If no personalization options exist, the chatbot asks the customer an open question, such as “How may I help?”.

Once the customer responds, the chatbot tries to interpret the request to determine intent and then decide what to do next. For example, if the customer replies with “I want to check my balance,” the chatbot would first identify and verify them before showing their balance.

Once the task finishes, the chatbot asks if the customer needs more help. The customer can respond by asking another question, requesting to chat with an advisor, or replying ‘no’. If the customer replies with ‘no’, the chatbot can offer a survey based on context.

If intent is not established or understood, the chatbot passes the customer to an advisor.

If the customer chooses to speak or chat with an agent and there is a long wait time or it is outside business hours, then the chatbot can present a suitable message.

The chatbot continues in this fashion, creating a conversational loop and building context between itself and the customer to better solve their query.

  1. A chat interaction is initiated (reactive or proactive) across a supported channel.
  2. The customer receives a standard welcome message from the chatbot.
  3. Customer information and/or context is retrieved from:
    • Customer profile information in External Contacts
    • API call to third-party data source
  4. The customer receives a personalized message or is handed over to an agent. Examples include:
    • Custom message or update: “Your next order is due to arrive on Thursday before 12.”
    • Customer is handed over directly to an agent because they owe an outstanding balance.
    • If the customer is not handed over to an agent, the customer could end their chat, confirm the contact reason, or continue.
  5. Assuming the customer has moved on from the Personalization stage, the interaction is sent to a chatbot (for example Genesys Dialog Engine) which asks an open-ended question like: “How may I help you?” to determine intent and capture the customer’s response.[BL1]
    • If intent and slots are returned, the conversation moves to the correct point in the interaction flow, for example;
      • Automated notification task (such as display balance)
      • Handoff to live agent
    • If intent and slots are not returned, the conversation returns to the interaction flow and the customer is handed off to an agent.
  6. Upon completion of a task,the interaction is sent to a chatbot (for example Genesys Dialog Engine) which asks a follow-up question like: “Is there anything else I can help you with?”
    • If the customer responds “yes,” they return to Step 5: “How may I help you?”
    • If the customer responds “no,” then the conversation returns to the interaction flow
    • If the customer responds with a more advanced answer, then determine intent and entities for further processing.
  7. Customer information and/or context is retrieved to determine whether to offer a survey.[BL2]
    • If a survey is offered, the interactions is sent to a chatbot.
    • If no survey is offered, the interaction flow shows a goodbye message and ends
  8. The survey is executed. The survey questions are configurable by the customer on a business-as-usual basis in the chatbot and therefore no dialog flow is defined here.
  9. The interaction flow presents a goodbye message and ends the chat

Business and distribution logic

Business Logic

NLU:

  • Intents: The goal of the interaction. For example, a “switch flight” intent returned by the NLU indicates that the customer receives a payment business process.
  • Slots: Additional pieces of key information returned by the NLU. These pieces can accelerate the conversation by prepopulating answers to subsequent questions.

BL1: Agent Handoff: The customer can ask to connect to an available agent. At that point, the chatbot disconnects and the chat transcript (excluding sensitive data) appears in the agent desktop.

BL2: Survey: The customer can determine whether to address a survey or not. This survey can be based on:

  • Customer profile information in External Contacts
  • Customer journey data
  • API call to third-party data source

User interface & reporting

Agent UI

Chat transcript between customer and chatbot is populated in the chat interaction window in the agent desktop.

Reporting

Real-time Reporting

With Genesys Cloud CX, you can do flow reporting and use flow outcomes to report on chatbot intents.

See the Flows Performance Summary view and use flow outcomes statistics to help you determine performance issues for specific chatbot flows, and gather data about self-service success. Use the chatbot flow data to improve outcomes.

Use the Flows Performance Detail view to see a breakdown of metrics by interval for a specific chatbot flow, and to see how chatbot interactions enter and leave a chat flow.

The Flow Outcomes Summary view displays statistics related to chats that enter Architect flows. These statistics can help you determine how well your chatbot flows serve customers and gather data about self-service success.

Historical Reporting

We are working on providing more chatbot reporting in the future, including building your own chatbot reports.

Customer-facing considerations

Interdependencies

All of the following required: At least one of the following required: Optional Exceptions

General assumptions

  • Handoff to agent is on the same channel.
  • The customer is responsible for the build of the natural language bot model and providing the bot training of utterances, intents, or slots. Professional Service may be engaged to develop the model.
  • Survey capabilities are provided by chatbot provider QA functionality (for example, Amazon Lex) and need customization.
  • Chatbot integration is not HIPAA-compliant.
  • Third-Party Chatbots are enabled via the Integrations Registry and informational through AppFoundry.
  • Customers use their own third-party Chatbot accounts for Integration Services.

Customer responsibilities

N/A

Related documentation

Document version

V 1.4.0 last updated November 9, 2021