# LivePerson Integration
Disclaimer: This document was last updated on 03/01/2020. If you have suggestions please contact us at training@clinc.com.
An integration between LivePerson (opens new window) and Clinc involves using Clinc as the NLU/NLP solution for any of the already pre-configured and supported frontend applications/platforms. Documentation that are useful for this section:
- General Developer Documentation (opens new window)
- Supported Channels and Features (opens new window)
- 3rd Party Bot Integration (opens new window)
- Supporting Rich Responses (opens new window)
# Development Overview
- Create a generic Bot and Bot User using the Live Person Bot Builder (opens new window).
- Create a third-party Bot Connector (opens new window) for custom integration.
a. Integrate Clinc with Live Person through Live Person Functions (opens new window)
b. Add a Function (FaaS) Integration (opens new window) to the bot. - Connect Bot to a third party platform (Apple Business Chat, Facebook Messenger, What's up. etc).
# OOB (Out-of-the-Box) Solution
Clinc’s OOB solution involving uploading a code package(linked below) to LivePerson Functions' built-in lambda integration service. This solution is designed to work without any configuration other than supplying runtime information like host, AI version, credentials, etc. The code package below contains inline documentation.
Download the JSON file here:
# Custom Lambda Integration
When creating your own lambda integration for LivePerson the following are requirements and considerations. It is also important to read the 3rd party bot integration documentation (opens new window) in the prior reference material.
- Making use of the API Key (App-Key) for authorization with the Clinc API.
- Handling conversation states such as close conversation and bot escalation to a live agent.
# Rich Responses
Read the Supporting Rich Responses (opens new window) reference material linked above.
# LivePerson Agent Integration/Escalation
When the LivePerson bot can no longer handle the end-user queries and is escalated to a live agent it is important for all relevant context to be maintained. The transcript and classification of each query will be immediately available to the live agent without any additional services through the Agent Escalation Summary Widget as long as the OOB solution is used and/or all considerations in the custom lambda integration were followed. What won't be available is any other standard response data or custom data provided in each response from Clinc. It is possible to build what LivePerson calls a Workspace Widget that has 2 possible functions. The first would be to fetch the previously explained extra user data and pass that information to the live agent. The second would be to make a query with each post-transfer user query to suggest a response.
# Usage Restrictions
Clinc integrates directly through LivePerson Functions (opens new window). These have a cap of 150,000 invocations a month, which are free of charge. Exceeding this cap without a mutually agreed upon and signed Order Form between the provider and Live Person will result in an inoperable service until that is rectified. Being proactive and getting an agreement in place before production usage is recommended.
# Dialog/Conversation State Management
This integration makes use of a Beta feature for dialog management called a Dialog Seed. The current dialog management requires passing in a unique identifier returned from the Clinc Query endpoint along with any subsequent query. This “dialog token” changes with each query and allows a conversation to maintain state. This new type of dialog management removes the ability to return to a previous state of conversation but allows setting a “Dialog Seed” that acts as a pointer to the most recent conversation state. This removes the need for any client-side state management as long as access to a unique identifier is available.
Last updated: 03/01/2020