Knowledge Retrieval

Why is Knowledge Retrieval useful?

Our Knowledge Retrieval feature performs intelligent semantic search across your personal document library to find exactly what you need. This eliminates the need for you to waste time trawling through all the documents you have, transforming hours of manual searching into seconds of precise information retrieval.

This page explains how to set up and use the Knowledge Retrieval feature to create intelligent document search capabilities within your workflows and agents.

How to enable Knowledge Retrieval?

Here are the steps to set up Knowledge Retrieval functionality in your workspace:

1. Navigate to the Library tab and click on the + Resources button on the top right corner

2. Select the document format and enter the details into the relevant text boxes, then click the Next button

3. Click on the upload box to upload your file, select the file you wish to upload, then click the Next button

4. Wait for the system to process your document, this converts it into searchable text format. Once done, click the Next button.

5. Edit and tweak the chunking strategy based on your preferences. This step is optional. Once done, click the Next button.

6. Review the document preview and click Next, then Confirm to add it to your library.

7. Go to the Workflows tab and click on the workflow you want to enhance with Knowledge Retrieval capabilities.

8. Click New Step and search for Knowledge Retrieval. Click on the All Actions tab and click on the Knowledge Retrieval box to add a Knowledge Retrieval node to your workflow.

9. Click on the new Knowledge Retrieval Node and select the dataset (document) you uploaded from the dropdown menu.

10. Configure the query settings. You can just enter Extract for basic text extraction. Finally, click Save.

How to use Knowledge Retrieval?

Here are some guidelines to make Knowledge Retrieval as useful as possible for your needs:

1. To ensure that your agent can interface with your Knowledge Retrieval workflow, go to the Agents tab and choose the agent you want to enhance with Knowledge Retrieval capabilities.

2. Click Edit Agent to access the agent's configuration settings.

3. To integrate your Knowledge Retrieval workflow into the agent:

   • Click on the Add button beside Workflows in the agent settings

   • Then click on the Add button beside the previously created Knowledge Retrieval workflow, then click on Add again to add the workflow

   • In the text box on the left, set up a simple system prompt like "Give me the prices and other information when I ask". Finally, hit the Save button

4. Example queries you can ask:

   • "What is the price of [specific item]?"

   • "What models do you have available?"

   • "Find information about [topic] in the documents"

   • "Extract details about [specific subject]"

5. The system will automatically:

   • Run semantic search across your document

   • Understand the meaning and context of your query (not just keywords)

   • Return relevant information with proper context

   • Provide natural language responses based on document content

6. You can ask follow-up questions in natural language, and the system will continue to search and provide relevant information from your uploaded documents.

Benefits

• Instant Information Access: Find specific details in seconds instead of hours of manual searching

• Semantic Understanding: AI grasps meaning and context, not just keyword matching

• Natural Language Queries: Ask questions in everyday language without technical syntax or coding

• Comprehensive Search: Search across multiple document formats simultaneously

Use Cases

• Legal Team: Find specific contract clauses across all your legal documents

• Support Team: Locate troubleshooting steps from all your technical manuals

• HR Manager: Extract policy information from all your employee handbooks

Knowledge Retrieval transforms your document library into an intelligent, searchable knowledge base that understands context and meaning, making information discovery effortless and precise.