Managing Points

Managing Points for AI Integration

To effectively manage AI responses to specific messages, we can set up managing points across various platforms:

• Front: Set the inbox as the managing point.

• Gorgias & Zendesk: Utilize tags as managing points.

• Intercom: Both the team inbox and tags can be managing points.

Configuring AI Response Rules

If a customer wants AI to respond only to certain emails or channels, they can implement the following steps:

1. Set a Rule on the Agent Platform: Create a rule to tag messages, such as those originating from a specific email address.

2. Add the Tag to the AI Gateway: Configure the AI gateway with the tag as a managing point.

Example: Gorgias Configuration

We can use rules in Gorgias to tag tickets based on factors like the message channel, sender's email, ticket status, and more. These tags can then be included as part of the AI's managing points. For example, you can tag messages originating from specific channels and configure the AI to respond only to those messages. This allows for more targeted and efficient management of support tickets.

1. Create a Tag Rule:

   • Set up a rule in Gorgias to tag incoming messages based on criteria like message channel

     

2. Then log into aissist console and add the tag into gateway managing point list

3. You can also add tags into AI's stop list, so AI won't respond to ticket with those tags.

   

Example: Zendesk Configuration

Please note that in our Zendesk gateway setup, there are two distinct integrations. One integration allows our AI to manage Zendesk tickets (from the email channel), while the other enables AI to handle conversations (from the Messaging channel). These separate integrations are necessary because Zendesk operates two systems that rely on different sets of APIs.

For tickets from the email channel, we can configure management points similar to the previous Gorgias example. However, setting up management points for conversations is more restricted. This limitation arises because Zendesk does not provide access to conversation tags, ticket IDs, or other useful metadata through the Conversation API.

If you want the AI to handle only specific Zendesk conversations, we can use the Smart Tag feature. For example, in your Zendesk setup, you can configure the Messaging system to collect customer names, emails, or other details, and then send a follow-up message as illustrated in the image below.

In the Aissist console, you can set up smart tags to determine which messages the AI should or should not handle. Here are some examples:

1. Respond Only to Tagged Messages: Enable the "Respond only to tagged" option and define a tag such as "message_ai_can_reply". For instance, in the scenario configuration, you can specify a condition like "email domain is aissist.io". When a customer's email domain matches aissist.io, the AI will apply the tag "message_ai_can_reply" and proceed to reply to the message.

   
2. Exclude Certain Messages: Define a tag like "message_ai_cannot_reply" and, in the scenario settings, specify conditions such as "email domain is test.com". Additionally, select the options "No response for current interaction" and "No response after current interaction". This ensures that if a customer's email domain is test.com, the AI will apply the tag "message_ai_cannot_reply" and refrain from responding.

3. Topic-Specific Responses: Create tags for specific topics, such as "Pricing", "Order_Tracking", etc. After the user describes their issue, if the scenario matches one of these tags, the AI will respond accordingly.

These configurations allow precise control over which messages are handled by the AI.