AI Resolutions Log

Overview

AI Resolutions Log report lets you see all inbound and outbound messages happening in AI Chat. This is a key tool for enhancing content.

For Accounts on the Upgraded Context LLM Platform or in the Upgrade Process

Unlock the power of this report to uncover end user inputs resulting in "no_response". These instances indicate when our AI Chat didn't find a suitable answer to a user query. By analyzing trends in "no_response" response name field, you gain valuable insights into requests that need attention. Create and refine new responses to enhance user confidence and satisfaction, driving better engagement and outcomes.

This report replaces the following reports:

• NLP Input Response Analytics report

• NLP traffic Log

Report Components

Component	Definition
Page Name	Name of the specific chat page
Page Type	Channel where the conversation was recorded.
Conversation Code	All messages from a given user on a given date (local time), on a given bot page.
End User Code	Unique users. A customer who holds conversations on multiple dates is still a unique end-user.
Message Code	Unique code of the message
Message Type	End User Button: Identifies if a user clicked on a button that continued the conversation flow End User Quick Reply: Identifies if a user clicked on a quick reply that continued the conversation flow Identifies if a user provided a quick reply, for example, a number to continue the conversation flow Bot Response: automatic response made by virtual assistant End User Text: text input sent by the end user
Message Text	A specific message sent by either party
Intent	The purpose or goal behind the end user's message. If Intent = No Match, it means that the platform didn't identify the Intent of a question and didn't respond to a user. No Match Intent is only relevant to clients who have not yet upgraded to the Context LLM Platform.
Intent Category	The broad classification or theme of the user's intent
Intent Sub Category	A more specific classification or subset within an intent category.
Response	The answer provided by the AI Chat.
Response Name	The identifier or label assigned to a specific response. If Response Name = No Response, it means that the platform didn't provide a response to a user. If Response Name = context_llm_generated, it means that the response was generated by LLM.
Is Generative Response	If IsGenerativeResponse = TRUE, it means that the response was generated by scraping web import & documents If IsGenerativeResponse = FALSE, it means that the response is prewritten, a response function, or gathered from the feed.

Tips

For Accounts on the Upgraded Context LLM Platform or in the Upgrade Process

Search by ''no_response'' Response Name to identify topics without a content. Use the Dashboard to create new responses.

For Accounts That Have Not Yet Started The Upgrade Process

Search for No Match intent. It means that the platform didn't identify the intent of a question and didn't respond to a user.

1. Filter by Intent Category to get a better understanding of a specific Category.

   1. Ticketing

   2. Communications

   3. FB

   4. Parking

   5. Activities

   6. Health

   7. On-site

2. Monitor unique user interactions over time to identify patterns in engagement and tailor responses accordingly. Use unique identifiers to personalize interactions and track individual user journeys.

3. Filter by a conversation code to see a given conversation on a given date and time.

4. Determine which channels and pages are most effective for engaging users and driving conversions.

5. Categorize intents to better organize and analyze conversation data. Identify common themes and topics to optimize response strategies.

6. Track response performance metrics such as Response Name, Response Language, Content Type, and Content Group. Analyze which types of responses are most effective in driving desired outcomes and adjust content strategies accordingly.

7. Review specific message content to understand user inquiries, concerns, or feedback. Use sentiment analysis and keyword tracking to identify trends and address common user issues proactively.