graph TD %%{init: {'theme': 'mc','layout': 'elk'}}%% ParseData-w4yzj[<div><img alt="logo" src="/_astro/braces.Djq0PW4_.svg" style="height: 20px !important;width: 20px !important"/></div>Parse Data] style ParseData-w4yzj stroke:#a170ff Prompt-8bk4w[<div><img alt="logo" src="/_astro/square-terminal.BMOXc-nZ.svg" style="height: 20px !important;width: 20px !important"/></div>Prompt] style Prompt-8bk4w stroke:#a170ff ChatOutput-1d6fq[<div><img alt="logo" src="/_astro/messages-square.BaSDmT6g.svg" style="height: 20px !important;width: 20px !important"/></div>Chat Output] style ChatOutput-1d6fq stroke:#a170ff OpenAIModel-1ojkm[<div><img alt="logo" src="/_astro/openAI.BhmuxEs3.svg" style="height: 20px !important;width: 20px !important"/></div>OpenAI] style OpenAIModel-1ojkm stroke:#a170ff OutlookEmailComponent-co5vi[Email] style OutlookEmailComponent-co5vi stroke:#a170ff ParseData-w4yzj -.- Prompt-8bk4w linkStyle 0 stroke:#a170ff Prompt-8bk4w -.- OpenAIModel-1ojkm linkStyle 1 stroke:#a170ff OpenAIModel-1ojkm -.- ChatOutput-1d6fq linkStyle 2 stroke:#a170ff OutlookEmailComponent-co5vi -.- ParseData-w4yzj linkStyle 3 stroke:#a170ff

Outlook Inbox Synthesizer

🧩 Overview

This workflow automatically connects to a Microsoft Outlook inbox, extracts the content of recent emails, and generates a concise summary of unread messages. By leveraging an AI language model, it groups similar emails, highlights key topics, and presents the results in a chat‑style interface, providing a quick, human‑readable digest of a busy inbox.

⚙️ Main Features

Retrieves emails from a specified Outlook folder.
Converts email metadata and body into plain‑text format.
Builds a dynamic prompt that instructs the AI to summarize unread emails.
Generates a concise, professional summary using an OpenAI language model.
Presents the final summary as a chat message within the platform.

🔄 Workflow Steps

Component Name	Role in the Workflow	Key Inputs	Key Outputs
Outlook Email	Connects to Outlook and fetches recent messages from a chosen folder (e.g., Inbox).	Outlook folder selection, operation type (Get Message), maximum number of emails to retrieve.	Raw email data (structured with metadata).
Parse Data	Formats the raw email data into a readable text block according to a defined template.	Email data, template string, separator, result key, optional metadata mapping.	Formatted email text (used for prompting).
Prompt	Constructs a prompt that instructs the AI to summarize unread emails, including grouping and highlighting urgent topics.	Formatted email text.	Prompt message ready for the language model.
OpenAI Model	Sends the prompt to an OpenAI language model and returns the generated summary.	Prompt message, model configuration (name, temperature, max tokens, etc.).	Summarized text output from the AI.
Chat Output	Displays the AI’s summary as a chat message within the platform.	Summarized text, conversation and session identifiers.	Chat message visible to the user.

🧠 Notes

The Outlook connector must be authenticated and granted read access to the target folder.
The workflow is configured to operate on the latest 10 emails, but this limit can be adjusted via the max_count parameter.
Only unread emails are considered in the prompt’s instruction; already‑read messages are ignored.
The AI model is set to a conservative temperature (0.1) to ensure consistent, concise outputs.
JSON mode is disabled; the prompt includes the word “JSON” only if a schema is explicitly supplied.
The system assumes email bodies are plain text or UTF‑8 encoded; HTML content is ignored unless explicitly parsed.
If the inbox contains very large attachments, the workflow currently ignores them; attachment processing can be enabled if needed.
Rate limits of the OpenAI API and Outlook’s Graph API may affect throughput; throttling can be handled by adjusting the maximum token count or the number of emails processed.