Ctrlk
For the complete documentation index, see llms.txt. This page is also available as Markdown.

How to setup AI Agent (Step by Step Guide)

🔧 Overview

Kai is your AI Agent built into the klink.cloud platform to automate customer conversations across messaging channels like WhatsApp, Facebook, Instagram, LINE, and website live chat.

This guide walks you through setting up a fully functional Kai AI Agent—from knowledge base creation to deployment on messaging channels.

🛠️ Step 1: Create a New AI Agent

  1. Go to the AI Agent section.

  2. Click “Create New Agent”.

  3. Give your agent a name (e.g., “Escape Bar Reservation Agent”).

  4. Set Activity Timeout (default: 10 minutes). This means a conversation will automatically close if inactive for 10 minutes.

  5. Click “Create” and then go to the Studio to configure your agent.

📚 Step 2: Create a Knowledge Base

  1. Navigate to Knowledge Base tab.

  2. Click “Create New Knowledge Base”.

  3. Choose one of three training sources:

    • Write your own Q&A

    • Upload PDF documents

    • Sync website content via URL

Example:

To sync with a website:

  • Choose “Website URL”

  • Paste the business website (e.g., https://escapebangkok.com)

  • Select:

    • Crawl single page (for 1 URL), or

    • Crawl all pages under the domain

  • Click “Sync” and wait for training to complete.

🤖 Step 3: Configure AI Agent Behavior in Studio

  1. In the Studio, click “Add Agent”.

  2. Select the type of agent you want:

    • Example: “Reservation Agent”

      • Description: Handles reservations from 5PM to 1AM

      • Add happy hour info if needed

  3. Write a prompt like:

    “You are a reservation agent for Escape Bar, open from 5PM–1AM. Help customers book tables and inform them about happy hour from 6–8PM.”

  4. Set fallback or transition rules:

    • For example, if a customer asks about “menu” or “wine”, transfer the conversation to the Knowledge Base Agent.

🔄 Step 4: Create Agent Transitions (Optional but Recommended)

  1. In the Studio, click “Transition”.

  2. Add conditions like:

  1. Select the appropriate Knowledge Base you created earlier.

  2. Connect agents with drag-and-drop to visualize the flow.

🧪 Step 5: Test the AI Agent (Emulator)

  1. Use the built-in Emulator to simulate conversations.

  2. Try different messages like:

    • “Can I reserve a table for next Friday at 7PM?”

    • “Do you have a wine menu?”

  3. The AI will respond based on the configured prompt and transition to Knowledge Agent if needed.

🌐 Step 6: Connect Messaging Channels

  1. Go to Settings > Messaging Channel.

  2. Choose from:

    • WhatsApp

    • Facebook Messenger

    • Instagram

    • LINE

    • Website Live Chat

  3. For WhatsApp:

    • Simply scan the QR code to connect

    • No complex WhatsApp Business API setup needed

⚙️ Step 7: Set Workflow Trigger

  1. Go to Workflow Automation.

  2. Click Create New Workflow.

  3. Add Trigger:

    • When thread is opened

  4. Add Condition:

    • If channel = [LINE, Facebook, WhatsApp, etc.]

  5. Action:

    • Assign to AI Agent → Select your created agent

  6. Enable and save the workflow.

✅ Final Test

Send a real message through the connected channel and check:

  • Does the AI respond correctly?

  • Does it switch to the knowledge base agent when appropriate?

  • Are all reservation details correctly captured?

🎉 You’re Done!

You’ve now successfully built and deployed an Omni-Channel AI Agent with:

  • Reservation capabilities

  • Website-trained knowledge

  • Multi-channel deployment

  • Automated workflows

For questions or setup support, contact our team or book a 1-on-1 session.

Start building smarter customer conversations today with Kai AI Agent on klink.cloud.

Last updated