Ticket suggestion and insult detection using external API

Title: having a way, optional or not for an agent, to send his answer to a filter before sending it for real to the end user.

What is the idea or pain point:
linking an api like GPT4 or GrammarGPT would allow the said API to suggest change to tickets and/or warn manager of issues.

For example we could send the ticket answer to chatGPT so it can correct gramatical and spelling issues.
Also if the content is deemed hatefull if an agent is having a problem with a customer it will allow the system to warn a manager to monitor this ticket. This could be usefull in receiving a ticket to see if a customer is angry in the tone or has insults.

The new AI can really help in spelling/gramatical issue a lot better than integrated firefox spelling help.

Also the filter could refuse to send the ticket if more than a threshold of 10 spelling/grammar issues are detected etc…

Also the detection of insults and other harsh tone can help manager be alerted quicker of an issue with a customer.

Other possibilities are possible. system filters seems one way to add this but i cannot find any information on how to add one.


Expectation (not solution):

Having calls on ticket submission or answers that can send content to an API and have action taken based on the API return.

On agent submission of an answer , allow a call to an api that present a suggestion of a corrected response to send, then the agent can choose to send the original, the corrected one or to cancel the send action so the user can edit the message and make correction himself.

1 Like

I guess, you could try to set this up with www.make.com

Zammad can use Webhooks so you could use a Webhook as a trigger and then sent the message to GPT4.