Creating a Knowledge Search Widget

  • Updated

Get Enterprise AI+

Do you want access to this new feature? Contact your Zingtree Account Manager or our Support Team.


Search your Organization’s Knowledge from a single place.

Our Knowledge Search delivers quick and relevant responses, searching across all of your Organization’s sources of data in a single interaction.  

  • Reduce agent handle time when they don’t need to traverse multiple applications or stores of knowledge to find what they need
  • Deflect support calls by providing customers with access to accurate information on their own, quickly. 

Benefits of Semantic Search

  • Accuracy: Semantic search provides more accurate results because it understands your intent, not just keywords.
  • Saves Time: It helps you find what you're looking for faster by presenting the most relevant results upfront.
  • No Need to Be a Search Pro: You don't have to be an expert at crafting perfect queries; semantic search does the heavy lifting.

How does it work?

Traditional vs. Semantic Search

Traditional search engines match keywords in your query with keywords in the content, by finding those exact word pairings. This approach can be limiting because it may not find content that uses similar phrases or understands natural language.


Semantic search, on the other hand, goes beyond keywords. It understands the meaning behind your query. It uses your contextual knowledge, and understands the user intent and related materials to deliver the most relevant results.


AI-driven Semantic Search

  • Large language models play a crucial role in powering AI-driven semantic search. These models are designed to understand and generate human-like text based on the input they receive. 
  • Query Understanding: When you enter a search query, large language models are used to understand the query better. They can break down the query into its constituent words and phrases, analyze the relationships between them, and identify the intent behind your search.
  • Contextual Analysis: These models excel at understanding context. They consider the words you've used and their order to determine the context of your query. 
  • Semantic Understanding: Large language models are trained on massive amounts of text data, which enables them to understand the semantics of words and phrases. They can identify synonyms, antonyms, and related terms. 
  • Generating Relevant Results: After understanding your query, these models can generate search queries that are more refined and tailored to what you're looking for. They can expand your query to include synonyms or related terms, increasing the chances of finding relevant results.

  • Ranking and Personalization: Large language models also play a role in ranking search results. They consider your query and the context to determine which results are most likely to satisfy your specific needs. This personalization helps ensure that the top results are the ones most relevant to you.


You’ll find access to this through the Knowledge Search page in the top navigation bar.

Note: AI Knowledge Search is in alpha, which means it’s still a work in progress and may not be available to you yet.


Here, you can create new search widgets or manage existing ones. Create different widgets for different audiences within your Organization.

Create a new search widget:

  1. Select Add Search Widget:

  2. Select your widget Type based on the audience you intend for the widget: Customer Self-help, Agent Support or Internal Process:

  3. Use the search widget configuration to customize and create your search. More details below.

Configure source content

Define the content your users will be searching for. Your search widget can search Zingtree decision trees as well as external sources like Zendesk Guides, Confluence, and Salesforce Knowledge. Many more coming soon.

First, you’ll need to determine the knowledge sources you want your search to reference.

  • Determine which decision trees (if any) from your Organization should be referenced in your search
    • Trees need to be published to be accessed by your search.
    • Tree tags will manage access to trees in search for those search widgets relying on agent credentials or SSO for authentication. 
      • Only trees with tags matching those assigned to the agent will be included in the search index and results.
      • Agents with no tags assigned will have access to all results.
  • Determine any external sources you want your search to reference

A search widget must contain at least one source.



Define where your users can find your search widget.

Our Knowledge Search widgets are designed to be flexible – so you can deploy it where your users need it the most.

Knowledge Search widgets can be embedded in a website or application, it can appear right in line with a decision tree, it can work inside an integration like Salesforce, or it can function as a standalone page for agents to start their interactions. It can serve as a knowledge hub and starting point – or serve as an assistant to a workflow when the conversation takes an unexpected turn.

Access to search data requires authentication


Here’s what to know about each:

Deployment: Standalone

Your search can be accessed as a single-page application, with a unique URL. 

  • Offered for Agent Support and Internal Process
  • Select a title for your Search Page

Authentication: Access can be managed through Zingtree Agent Credentials or SSO


Deployment: Embed in a web page or app

You can embed your search into an existing web page or application. 

Authentication is required via API key. See these instructions for full details.


Deployment: In-Tree Search Widget

You can add your search widget to your decision trees, where users can search without leaving the tree or interrupting their session. This displays as a persistent search icon at the top or bottom of each node on your tree.

  • Offered for Customer Self-Help, Agent Support and Internal Process 
  • Select which trees you want the search to be available
  • Select where you want to display the search access: top or bottom of the tree


  • Access can be managed through Zingtree Agent Credentials or SSO. Or through an API key
  • The selection must match the security configuration of the tree where the search widget appears
    • For example – If your trees require SSO, then your search widget should also require this.
    • If the authentication methods do not match and cannot be passed from the tree into the search, then the search widget will not appear for the user.
    • Access to search data requires authentication, so publicly accessible trees will not support the search widget.

Deployment: Salesforce Integration

You can add search to your Zingtree integration within Salesforce, where it will display as an additional tab for agents to access alongside their decision trees.

  • Offered for Agent Support and Internal Process.
  • Use the unique search widget ID provided.

See this page for app package installation and setup instructions

Content Scope

Customize your search experience using contextual data and content segments, narrowing the scope of the index to deliver more accurate and catered results.

  • Context-Aware: Show content relevant to the user's location - e.g. current webpage or case record type.
  • Audience targeting: Show content relevant to the user’s profile - e.g. support tier, product access, or support line.

Contextual data is then passed into your search widget through URL parameters.

See full setup details for Content Scope

Help Us Improve

Do you or your users have feedback about your search results? Use the thumbs up and thumbs down as you’re using your search widget to help improve it.

Was this article helpful?

0 out of 0 found this helpful

Have more questions? Submit a request

Didn't find what you need?

Our friendly customer support team is here to help

Submit a Ticket

Looking for help or advice?

Reach out to our knowledgeable community of users.

Zingtree Community