Notes

Hi, good morning everyone!

I'm going talk about an automated email response workflow called eBuddy that aims to be a friendly assistant to the Comms DRIs.. It takes an incoming email, processes it, and drafts a tailored response with minimal human effort.


-----------------------Slide 2----------------------------

When an email arrives, a Mail rule triggers an AppleScript that extracts emails details and encodes it into a JSON and sends it to the Python script which is the core engine of the workflow. 

It takes the parsed email data and drives all the logic and decision-making for the automation

It first decodes the JSON to extract the key email fields — like sender, subject, message content, and thread metadata and then sends this information to Apple-Genai. 

GenAI analyzes the message and extracts key search terms, which are then used to run targeted queries against Confluence.

The workflow is also designed to handle cases where there’s no good documentation match or the sender isn’t asking for docs. In those cases, the language model is still able to generate an appropriate response based on the email alone.


Confluence returns the top two pages for each query— giving us enough context.

Those pages are then passed back to GenA to draft the email content which would tailored to the original message, and sent back to the Python script.

With the drafted email content ready, we then trigger an AppleScript to create the actual email draft. 

This script generates a ‘Reply All’ message within the original thread to preserve context and continuity. The draft can then be reviewed by the comms DRI before sending. This ensures that while the process is highly automated, there’s still a human in the loop — providing oversight and maintaining quality before any response goes out.

-----------------------Slide 3----------------------------
Demo 

Now that we’ve seen how the automation works under the hood, let me show you a live example so you can see it in action end-to-end.

The rule for this demo is set up such that any email sent to me will automatically be processed for a draft response but I have also created a script that will trigger a draft response for any selected email while showing the logs 

1. So let's go over that first --- [ long thread email ]


2 . Now let's see how this might look when we receive an email [ send myself an email ] 




The Mail rule has just triggered AppleScript, which converted the email into structured JSON. That was picked up by the Python script, which decoded the fields and passed the message to GenAI.”

“GenAI extracted search terms, sent three queries to Confluence, and is now using the top six pages to draft the response content.”

And here’s the final draft — automatically created as a ‘Reply All’

-----------------------Slide 4----------------------------

Key Advantages of This Automation:

1. Significant Time Savings
It reduces the manual effort involved in reading, researching, and drafting replies — especially for repeatable or documentation-related inquiries.

2. Operational Efficiency

3. Human Oversight Maintained

4. Scalability
This solution can easily scale with increased email volume, without requiring a linear increase in effort.

5. Adaptability to Non-Doc Queries


6. Configurable Across Use Cases
“This workflow isn’t limited to comms-related emails. It’s designed to be highly pluggable and configurable. Different AppleScript rules can be easily set up to handle other categories of emails — across teams, topics, or mailboxes. That means this same framework can be extended to support multiple use cases with minimal changes.”




------

Demo