Using ThinkAutomation With ChatGPT
The latest release of ThinkAutomation includes a ChatGPT action.
One of the uses of ChatGPT with ThinkAutomation is classifying emails (or any text).
You can send a prompt to ChatGPT and assign the response to a variable. You can then use the response further in your Automation. You can use ChatGPT to summarize text, to classify emails, to extract information from unstructured text and much more. ThinkAutomation automatically handles OpenAI rate limit responses.
For example, suppose we receive various incoming emails into a mailbox. Some will be genuine sales enquiries, some will be product support requests and some will be marketing emails. We need to classify these so they can be dealt with by the correct team.
- First, we send a prompt to ChatGPT with a summary of the email text.
- We ask ChatGPT to tell us if the text is 'sales', 'support' or 'marketing'. We give some additional information to help ChatGPT make the choice. ChatGPT is able to classify text using one of the standard models. You can improve accuracy even further by fine-tuning a model with your own training data.
- Based on the response from ChatGPT we request confirmation from a user. ThinkAutomation sends a validation request to a user and waits for the response.
- If the email is a genuine sales enquiry then we update our CRM system which also sends a notification to the sales team.
- If the email is a genuine support request we then create a support ticket.
- If it's a marketing email we simply move the email to a different folder.
chatgpt-example-email-automation.webp
email-classification-using-chatgpt.webp
| Files | ||
|---|---|---|
|
|
||
|
|