> For the complete documentation index, see [llms.txt](https://kinochat.gitbook.io/help-center/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://kinochat.gitbook.io/help-center/quick-start/enable-ai-agent.md). # Enable AI Agent AI Agent is an intelligent assistant for customers. It helps agents quickly find relevant information from your knowledge base, enabling faster and more accurate responses to customers. Go to **AI Agent** to learn more about the product, explore its capabilities, and view relevant performance data and benchmarks.
If you're interested in our AI products and would like to get started with our AI services, you can reach out to us by email or click **Contact Sales** to learn more.
Once you've purchased our AI services, you can follow the **Configuration Checklist** to set up your AI Agent.
Configure your AI Agent in the following 5 simple steps. #### Step 1: Add Knowledge Knowledge Base is the primary source of information used by AI Agent to answer customer questions. You can: * Click the **Configure** **button** in step 1 of the checklist * Go to: **AI Agent > Knowledge Base or Content > Knowledge Base** Before enabling your AI Agent, add the information your customers frequently ask about.
#### Create an Article Click Write Article in the top-right corner and fill in the following: * **Title & Content** * The main information used for search and AI responses * **Type** * FAQ – for structured question-and-answer content * Article – for longer or descriptive content * **Category** * Organize your knowledge into custom categories for easier management * **Language** * The language of the content * **Permissions** * Enable this option to allow AI Agent to access and use this content
#### Bulk Upload Articles Click Bulk Upload in the top-right corner. 1. **Download the template** 2. Fill in your content and save it as a CSV file 3. Click Upload Document to import the file After uploading, the system will display: * Number of successfully imported rows * Failed rows with error details #### Manage Knowledge All uploaded content will appear in the knowledge table. You can: * Search using the search bar * Filter by category or other attributes * Click a title to view and edit content #### Edit & Delete You can update or delete any knowledge entry at any time. ⚠️ Deleting content will immediately affect AI Agent responses and cannot be undone.
#### Step 2: Configure Agent Skills (Optional) Agent Skills define what your AI Agent can do. You can: * Click the **Configure** **button** in step 2 of the checklist * Go to: **AI Agent > Agent skills** KinoChat provides four core AI Skills:
1. **Fixed Answer:** Retrieves answers directly from the FAQ based on the agents' questions. 1. You can configure fallback strategies if no exact match is found, including fixed fallback message or still provide a possible answer 2. **AI Generated Answer:** Generates an answer by searching both FAQ and Article content. AI synthesizes the content to provide a relevant and coherent response. 1. You can configure fallback strategies if no exact match is found, including fixed fallback message or still provide a possible answer 3. **Free Talk:** Enables AI Agent to act as a conversational chatbot, handling casual or open-ended interactions. 1. Free Talk allows you to customize the AI’s behavior using prompts. You can define how the AI responds in open-ended conversations, including tone, style, and response guidelines. #### Step 3: Set Up Human Handoff (Optional) Human Routing determines: * When the AI Agent should transfer a conversation to a human agent * What happens after a transfer is triggered You can: * Click the **Configure** **button** in step 3 of the checklist * Go to: **AI Agent > Human Routing**
* **Routing node** * AI only: Conversations are handled entirely by AI without any human agent involvement. * AI first: Conversations are initially handled by AI and are automatically transferred to a human agent when predefined escalation conditions are met or the user requests human assistance. * Condition rule: Conversations are routed to a human agent when the user meets the configured conditions; otherwise, they are handled by AI first. * **Escalation rule** * Failed AI attempt: Automatically transfers the conversation to a human agent when the number of failed AI responses reaches the configured threshold. * Sentiment detection: Automatically transfers the conversation to a human agent when AI detects negative emotions, such as anger or frustration. * Business rule: Define custom business rules or scenarios that require the conversation to be transferred to a human agent. * **Transfer message:** Configure the message shown when a conversation is transferred to a human agent. #### Step 4: Configure AI Profile Start by defining how your AI Agent looks, communicates, and represents your business. You can: * Click the **Configure** **button** in step 4 of the checklist * Go to: **AI Agent > Deploy**
Manage AI Agent’s basic information, including: * Profile: Agent name, Agent description, Voice tone, Language * General context: Company name, Company description * Greeting message: Welcome message when chat starts * Quick reply buttons: Pre-set answer when customer click the label button * Handoff button: Configure the customer-facing button used to request human support. * Fallback message: The fallback message shown when a required AI service is unavailable, such as when Free Talk is unavailable Save the configuration. #### Step 5: Test & Accepting chat At this step, you've completed all the required configurations. You can: * Click the **Configure** **button** in step 4 of the checklist * Go to: **AI Agent > Deploy** Use the **Preview** to test real AI Agent's responses, inclouding Greeting Messages, Quick Replies, fallback responses, and Human Routing behavior. All conversations created in **Preview** are excluded from your customers' actual data records, including **Conversations** and **Dashboard**. You can test your configuration without saving and restart the Preview conversation at any time until you're satisfied with the results.
Once you've verified that your AI Agent is ready to go live and serve real customers. Return to the **AI Agent > Get Started** page and turn on the **Accepting chat** button to start handling customer conversations with AI. #### What's Next? After your AI Agent goes live, continue monitoring your customer conversations and improve your AI over time. We recommend regularly reviewing: * Update Knowledge * Manage Skills * Modify Human Handoff * Configure AI Profile * Test your Agent
#### **View conversations' performance** * **Conversation Trace** You can review **Conversation** to evaluate your AI Agent's responses to real customers and how each response was generated. Go to: **AI Agent > AI Conversation** The response details include three key components: 1. **Knowledge sources:** The articles referenced to generate the AI response. Click an article to view its details in the Knowledge Base. 2. **Workflow:** The workflow used to generate the AI response, including **Fixed Answer**, **AI Generated**, **Free Talk**, and **Transfer to Human**. 3. **Transfer reasons:** The reasons why a conversation was transferred to a human agent, including a customer request, reaching the failed response threshold, meeting business rules, or detecting negative customer sentiment.
* **Escalated to human** Review transferred conversations and identify cases that could be handled by AI in the future. Go to: **AI Agent > AI Conversation > Escalated to human**
* **Customer Feedback** Monitor customer feedback and satisfaction to understand how well your AI Agent is performing. Go to: **AI Agent > AI Conversation > AI involved**
* **AI Performance** Use your analytics and conversation data to identify opportunities to improve your AI Agent's accuracy, coverage, and resolution rate. Go to: **AI Agent > Overview**
Your AI Agent should be treated as an ongoing system that improves as you learn more about your customers and their questions. Regularly refine your knowledge, routing rules, and AI profile to deliver more accurate support. #### **Update Configuration & Publish Version** According to customer conversation feedback, identify where AI Agent can improve. Adjust settings such as Agent Skills and Human Routing to refine how your AI Agent responds to customer questions and handles conversations. Once you've finished configuring your AI Agent, you can review your setup on the **Get Started** or **Version Management**.
You can also click **Preview** to open the Preview panel and test your AI Agent's responses in real time. This helps you verify that your configuration changes are working as expected before publishing them.
Once you're satisfied with your changes, publish a new version to apply the updated configuration to live conversations. You can add a version note when publishing a new version to document why the changes were made, making it easier to review and track updates later.
To review a published version, click its name to view the configuration details for that version.