Written By William Bowen
Last updated 14 days ago
Intro
This is our knowledge base search agent! This intelligent agent seamlessly processes incoming messages from your customers by extracting questions and searching your selected knowledge base materials for relevant information, providing accurate, contextual responses based on your company's specific documentation.
For example, if a customer asks "How do I reset my password?", the agent will instantly locate and deliver the exact steps from your knowledge base articles.
The agent's power lies in its ability to understand complex queries and provide precise answers drawn directly from your knowledge base, such as explaining your platform's API integration process or detailing your subscription management procedures, ensuring consistent and accurate support that perfectly aligns with your company's documentation.
Use cases
Healthcare Equipment Provider
Description: Assists medical professionals with equipment specifications and operating procedures from the provider's knowledge base.
Example: When a nurse asks "How do I calibrate the XR-500 patient monitor?", the agent provides precise calibration steps from the equipment manual.Financial Services Company
Description: Helps customers understand financial products and services using the institution's documentation.
Example: When a client asks "What documents do I need for a mortgage pre-approval?", the agent lists all required documentation from the company's lending guidelines.Software Company
Description: Supports users with product functionality and customization using their software documentation.
Example: When a user asks "How do I create custom reports in your CRM?", the agent provides step-by-step instructions from the software manual.Manufacturing Company
Description: Assists distributors and customers with product specifications and technical details.
Example: When a distributor asks "What's the maximum load capacity for the M2000 conveyor belt?", the agent provides exact specifications from the product documentation.Educational Institution
Description: Helps students and faculty navigate administrative procedures and policies.
Example: When a student asks "What's the process for requesting a leave of absence?", the agent outlines the procedure from the institution's policy handbook.Retail Chain
Description: Supports staff with inventory management and store procedures.
Example: When an employee asks "How do I process a return for an online purchase?", the agent provides the return protocol from the operations manual.Insurance Company
Description: Guides agents and customers through policy information and claim procedures.
Example: When a policyholder asks "How do I file a claim for water damage?", the agent provides detailed claim filing steps from the claims handbook.Travel Agency
Description: Assists customers with booking policies and travel requirements.
Example: When a traveler asks "What's included in your luxury cruise packages?", the agent provides detailed package information from the travel catalog.Automotive Dealer
Description: Helps customers and staff with vehicle specifications and service information.
Example: When a customer asks "What's the maintenance schedule for a 2025 EV Model X?", the agent provides the exact service intervals from the manufacturer's documentation.Real Estate Agency
Description: Supports agents with property information and transaction procedures.
Example: When an agent asks "What documents are needed for a commercial lease agreement?", the agent provides the complete checklist from the agency's transaction manual.
Upload your knowledge
To get this agent running, upload your knowledge base docs to the knowledge base node!
Screenshots
Template
Example{
"$schema": "https://web-api.clerk.chat/pipeline-schema",
"name": "Answer questions from knowledge base",
"nodes": [
{
"type": "ai_bot",
"name": "Extract question",
"triggeredBy": [
"unreadMessage.fromAny"
],
"responseType": "json",
"nodeConfig": {
"modelProvider": "openai",
"modelVersion": "o3-mini",
"maxTokens": null,
"temperature": null,
"variables": {},
"prompt": "Your job is to read the conversation history, specifically the most recent messages and find the user's most recent unanswered question. This will be at the bottom of their messages. Output the user's most recent question.\n\nIf the user's most recent message isn't a question, then leave this output blank. \n\nAlso output your reason for why you chose to output what you did.",
"promptSections": [],
"sectionTemplates": {},
"responseSchema": {
"type": "object",
"required": [
"reason",
"question"
],
"properties": {
"reason": {
"type": "string"
},
"question": {
"type": "string"
}
}
},
"opts": {
"sendStructuredConvo": true
}
}
},
{
"type": "ai_bot",
"name": "Answer question",
"triggeredBy": [
"unreadMessage.fromAny"
],
"responseType": "user_message",
"nodeConfig": {
"modelProvider": "openai",
"modelVersion": null,
"maxTokens": null,
"temperature": null,
"variables": {},
"prompt": "Your job is to answer the user's question, conversationally. \n\nHere is the knowledge you can use to answer their question: \n\n{{content}} \n\nYour output should be maximum 30 words. \n\nKeep conversational, if there is no question, just have a conversation. Maybe ask a question. But make it natural.",
"promptSections": [],
"sectionTemplates": {},
"responseSchema": {
"type": "object",
"required": [],
"properties": {}
},
"opts": {
"sendStructuredConvo": true
}
}
},
{
"type": "tool_runner",
"name": "Knowledge base search",
"triggeredBy": [
"unreadMessage.fromAny"
],
"responseType": "json",
"nodeConfig": {
"type": "knowledge_base",
"config": {},
"enabledTools": [
"search"
],
"tools": {
"search": {
"paramValues": {
"query": {
"path": "question",
"type": "pipeline-variable"
},
"sourceIds": [
{
"type": "fixed",
"value": 2469
}
]
}
}
}
}
},
{
"type": "trigger",
"name": "Start",
"triggeredBy": [
"userMessage"
],
"responseType": "json"
}
],
"edges": [
{
"name": null,
"sourceNode": "Extract question",
"destinationNode": "Knowledge base search",
"sourceVariables": null,
"filters": []
},
{
"name": null,
"sourceNode": "Start",
"destinationNode": "Extract question",
"sourceVariables": null,
"filters": []
},
{
"name": null,
"sourceNode": "Knowledge base search",
"destinationNode": "Answer question",
"sourceVariables": null,
"filters": []
}
]
}