NEWSRelevance Lab makes a strategic majority investment in kaman.aiLearn more

kaman.ai

IntegrationsPricingTemplatesBlogPapersDocs

Documentation

Guides, use cases & API reference

  • Overview
    • Getting Started
    • Platform Overview
  • Features
    • Features Overview
    • AI Assistant
    • Workflow Automation
    • Intelligent Memory
    • Data Management
    • Universal Integrations
    • Communication Channels
    • Collective Agent Memory (CAML)
    • Security & Control
  • Use Cases Overview
  • Financial Services
  • Fraud Detection
  • Supply Chain
  • Technical Support
  • Software Development
  • Smart ETL
  • Data Governance
  • ESG Reporting
  • TAC Management
  • Reference
    • API Reference
  • Guides
    • Getting Started
    • Authentication
  • Endpoints
    • Workflows API
    • Tools API
    • KDL (Data Lake) API
    • OpenAI-Compatible API
    • A2A Protocol
    • Skills API
    • Knowledge Base (RAG) API
    • Communication Channels
    • DSL Import API
Release Notes

How to Create an Agent

This guide walks you through creating an AI agent from scratch using the Agent Builder.

Step 1: Open the Agent Builder

Navigate to Build > Agents in the sidebar, or click New Agent from the home dashboard.

You'll see the Agent Builder with 11 configuration tabs:

Agent Builder - Basic Info

Notice the two builder modes at the top:

  • Form — Step-by-step form to configure each aspect of your agent
  • AI Builder — Describe what you want and let AI generate the agent for you

Step 2: Fill in Basic Information

Fill in the required fields:

FieldRequiredDescription
Agent NameYesA clear, descriptive name (e.g., "Customer Support Agent")
DescriptionYesWhat your agent does. This helps users understand its purpose.
VersionYesSemantic version (e.g., 1.0.0)
CategoryYesCategory for discovery (e.g., "Support", "Sales", "Operations")
Image/IconNoUpload a JPG, PNG, GIF, SVG, or WebP image (up to 10MB)

Click Next to proceed.

Step 3: Configuration

Set up how your agent behaves:

  • System Instructions — The system prompt that defines your agent's personality, rules, and behavior. This is the most important field.
  • Model — Select the AI model (GPT-4, Claude, Gemini, etc.)
  • Temperature — Controls creativity (0 = precise, 1 = creative)
  • Tool Confirmation — Enable to require user approval before tools execute

Tip: Write detailed system instructions. Include the agent's role, what it should/shouldn't do, response format preferences, and any domain knowledge.

Step 4: Attach Tools & Workflows

Click the Tools & Workflows tab to give your agent capabilities:

  • Tools — Functions your agent can call (e.g., query a database, send an email, search the web)
  • Workflows — Multi-step automated processes your agent can trigger

To add a tool:

  1. Click Add Tool
  2. Search for available tools by name or description
  3. Select the tools you want
  4. Configure tool confirmation settings (optional)

Step 5: Add Knowledge Base (Optional)

Click Knowledge Base to give your agent domain knowledge:

  1. Click Add Knowledge Source
  2. Upload documents (PDF, DOCX, TXT, CSV)
  3. Or connect to an existing knowledge base

Your agent will use RAG (Retrieval Augmented Generation) to answer questions based on these documents.

Step 6: Configure Data Sources (Optional)

Click Data Sources to connect your agent to live data:

  • Connect to databases (PostgreSQL, MySQL, MongoDB, etc.)
  • Connect to SaaS tools (Salesforce, HubSpot, Google Sheets, etc.)
  • Your agent can query this data in real-time during conversations

Step 7: Preview and Publish

Click Preview (the last tab) to review your agent configuration:

  • Verify all settings look correct
  • Click Test (top right) to chat with your agent before publishing
  • Click Publish to make your agent available

Step 8: Chat with Your Agent

After publishing, you can chat with your agent:

Chat Interface

  • Type your message in the input box
  • Click the Send button (or press Enter)
  • Your agent will respond, potentially using tools and knowledge you configured
  • Use suggestion buttons for quick interactions
  • Click the Model dropdown to switch models mid-conversation
  • Use the attachment button to share files with your agent

Tips for Great Agents

  1. Be specific in system instructions — "You are a customer support agent for an e-commerce store" is better than "You help people"
  2. Choose the right model — GPT-4 and Claude for complex reasoning, faster models for simple tasks
  3. Start with fewer tools — Add tools incrementally and test each one
  4. Test thoroughly — Use the Test button before publishing
  5. Use templates — Browse Templates for pre-built agents you can customize

On this page

  • Step 1: Open the Agent Builder
  • Step 2: Fill in Basic Information
  • Step 3: Configuration
  • Step 4: Attach Tools & Workflows
  • Step 5: Add Knowledge Base (Optional)
  • Step 6: Configure Data Sources (Optional)
  • Step 7: Preview and Publish
  • Step 8: Chat with Your Agent
  • Tips for Great Agents