AI Knowledge Agent

This AI Knowledge Agent is capable of identifying information gaps from customer inquiries and creating documentation to improve self-service. This article gives a bird’s eye view of the key components of the agent and its operational flow. This agent automatically generates and updates knowledge base content in Salesforce and we are expanding support for more platforms.

Key Components

The key components of the AI Knowledge Management Agent are:

  • Identifier

  • Creator

  • Reviewer

Identifier

The Identifier component is connected with the customer’s platform and identifies opportunities for creating and updating the knowledge base. Currently, the Salesforce platform is supported and more platforms will be supported in the future.

The Identifier works by analyzing newly-created customer support tickets and based on its analysis, it suggests whether:

  • A new knowledge article needs to be created

  • No changes have to be made in the knowledge base

The Identifier does its analysis in real-time. As soon as a ticket is closed, the AI Knowledge Agent is triggered. After the Identifier completes its analysis, the next step is handled by either the Creator or the Reviser.

Creator

The Creator component can write:

  • Knowledge base articles, including FAQs , Q&A, issue-solution articles, troubleshooting articles, how-to guides, etc.

The Creator generates drafts after gathering all the requisite information from:

  • Resolved support tickets

During this step, the Creator determines which Salesforce record type should be used. This means that the Creator decides whether a new knowledge article needs to be created and what kind of knowledge article needs to be created.

While generating drafts, Creator ensures that:

  • All the text fields and meta data fields are duly filled

  • The text is clear and meets the brand tone

  • Cross references are inserted wherever needed

The draft is then passed on to the Reviewer.

Reviewer

The Reviewer component analyzes the newly-created or updated knowledge articles for a set of predefined parameters and assigns them a score. Its job is to ensure that the content is accurate, complete, compliant, and clear.

Note: If the score is greater than the threshold value set by the AI Knowledge Management Agent admin, then the article is moved for publishing. Otherwise, it is sent back to the Creator.

The Reviewer is capable of analyzing knowledge articles on the following six parameters and assigning a score for each parameter:

  • Uniqueness

  • Completeness

  • Title Accuracy

  • Content Readability

  • Links Validity

  • Metadata Accuracy

It also checks the content on:

  • Company style guides

  • Brand guidelines

  • Legal guidelines

  • Compliance guidelines

  • Any other checks a company wants

Note: If a knowledge article is sent back to the Reviewer for the second time, then the Reviewer sends the article to a human (agent). This approach avoids endless review-creation-review loops.

Set up an AI Knowledge Management Agent

Setting up an AI Knowledge Management Agent consists of the following steps:

Once the agent has been deployed, you can view the Analytics for it in Knowledge AI Analytics.