Qontak presents the AI Agent feature to help businesses run smarter and more efficient automation. This feature not only allows chatbots to answer customer questions but can also perform various business actions directly within a single conversation flow.
With AI Agent, chatbots can assist in processes such as creating tickets, managing deals, and executing specific workflows without manual intervention from agents. This makes operational processes faster and more efficient while reducing repetitive workload on the team.
In this guide, you will learn the steps to create, configure, and use AI Agent in Mekari Qontak completely and structurally.
A. How to Create an AI Agent
- Log in to your Qontak account, then select the Chatbot menu.
- Then select the “AI agents” tab.
-
Next, on the AI Agents page, click “Create AI agent”.
You can also access the Create AI agent page through the Tree Diagram by following these steps:
1. On the Tree Diagram page, click the ‘add icon’, then select “AI agent”.
2. Then click on the following dropdown Select AI Agent. Then click “+Create new”. - You will then be directed to the following AI agent creation page. Fill in the AI agent name and AI Agent description fields. Then click “Continue”.
- The following screen will appear. Please wait until the system finishes preparing your AI Agent.
Next, you need to configure the Identity, AI resources, and Actions sections. In AI Agent configuration, the Identity, AI Resources, and Actions sections are optional. However, the AI Prompt is mandatory because it is used to guide how the AI Agent works.
In the next step, configure the Identity section. Here, users need to add an AI Prompt containing the goals (goals) and behavior (behavior) of the AI Agent, select the Tone of Voice for the AI Agent's communication style, and set the Idle Rules.
In the AI prompt field, the description of the AI Agent you previously entered will be automatically generated. Here, you can still edit the AI prompt field.
Then determine the Tone of voice or how the AI speaks when interacting with users.
-
Next, set the Idle Action, which is the action the AI Agent will take when the customer does not respond within a certain period, such as resolving the conversation (Resolved conversation), forwarding to an agent (Assign conversation), or following up (Follow up).
Below are some conditions that occur if you select one of the available Idle actions.
| Idle action | Preview | Actions to be taken |
| Resolve conversation |
|
|
| Assign conversation |
|
|
| Follow up |
|
-
Next, click the “AI resources” tab to configure knowledge sources that the AI can learn from. Here, you can choose External URL, File upload, or Text content as knowledge sources by clicking the ‘add icon’ on one of the AI resources.
- View when you select External URL as the AI knowledge source.
- View when you select File upload as the AI knowledge source.
- View when you select Text content as the AI knowledge source. Next, click the “Actions” tab to specify the actions the AI Agent can perform (not just answering chat). Click “Add actions” to proceed.
Currently, the available action is only API Integration. Use this feature to create actions connected to other systems for the AI Agent. Click “API Integrations”.
Then complete the information in each field below, then click “Save”.
Then click “Save” again to save all AI Agent data.
After that, the AI Agent will be displayed on the AI Agents page.
Additionally, the AI Agent will also be displayed on the Bot response page.
B. How to Configure API Integration on AI Agent Actions
Important
Make sure you have created the integration first through the API Integration menu in Chatbot Settings. Then select API Credentials to start configuring the action.
Go to the Chatbot menu.
Then select the Chatbot settings tab and choose “API integration”.
Next, click “Add” and select “API credential”.
Add the required credentials. Qontak will automatically encrypt the data to ensure the credentials remain secure. Fill in the fields below, then click “Save”.
After completing this, you can start configuring the API Integration tools.
C. How to Implement AI Agent on Tree Diagram
Follow these steps to integrate the AI Agent into an existing conversation flow:
- Go to the Chatbot menu.
- Then select the “Conversations” tab.
- Next, select one of the created Conversations.
On the following Tree Diagram, click the ‘add icon’ then select “AI agent”.
Select the AI Agent you have created.
- Then click “Save”.
- Click “Add condition”.
-
The Exit condition panel will appear. Below is an example of a condition created.
Define the flow for various possible results:
- Success: Connect to the Closing node when the goal has been achieved.
- Escalation: Direct to Assign Agent if the user requests human assistance or shows negative sentiment.
- Failure: Use the Generic Apology message if there is a timeout or API error.
- Preview & Publish: Use the Bot Preview feature to test the conversation flow and ensure variables work correctly.Important
Tips for implementing AI Agent:
- Enable the Listen Time feature to allow the AI Agent to wait until the entire context of the customer's question is received, especially if messages are sent in several chat bubbles.
- Note that the AI Agent may take longer to respond due to processing prompts, knowledge, or actions.
- On the WhatsApp channel, a typing indicator (typing indicator) is available so customers feel that the AI Agent is actively handling them. - Then click “Save”.
The Exit condition will then appear on the following Tree Diagram.
That is how to create, configure, and use AI Agent. Next, learn about how to compose effective prompts and Chatbot instructions here.