This is a submission for the Postmark Challenge: Inbox Innovators.

What I Built

InboxOps is a powerful AI-augmented operations platform that transforms inbound emails into real-time dashboards — using Postmark's inbound email parsing as its foundation.

• A customer emailed [email protected]: "My card was charged twice."
• An employee sent to [email protected]: "I need leave from June 15."
• A vendor mailed [email protected]: "Attached is this month’s invoice."
• A manager forwarded to [email protected]: "Approve $3,000 budget?"

And all those inboxes are

• Fragmented
• Human-dependent

People were copy-pasting from Emails into Tickets. Approvals were lost in threads. Support issues were never tagged. There was no single dashboard to see anything.

Inboxops automate workflows from:

• 📦 Customer Orders
• ✅ Approval Requests
• 🛠️ Support Tickets
• 🧑‍💼 HR / Leave Requests
• 🙋 Customer Enquiries

InboxOps turns every structured or unstructured email into actionable items using Postmark + LLMs, and visualizes them in a modern, modular dashboard, and helps create APIs to use in Internal Dashboards.

Single inbox automatically:

• Understand the intent (order, complaint, HR, etc.)
• Summarize the message using AI
• Assign tags and priority levels
• Route to the right internal dashboard
• Trigger concerned workflows

Demo

https://inboxops.onrender.com

Code Repository

https://github.com/DYNAMICMORTAL/inboxops

Screenshots

Order information, items, quantity, price extracted from Email Content

Customer enquiries, support, issues, requests, tickets are assigned Ticket ID concerned

Customer support tickets management

How I Built It

• 🟧 FastAPI (Python) — REST APIs and webhook handling
• 📬 Postmark — Inbound webhook parsing + email confirmation sending
• 🧠 Gemini 2.0 Flash — Summary, classification, tag + urgency prediction
• 🎨 Jinja2 + TailwindCSS - UI components, dashboards, modals
• 💾 SQLite + SQLAlchemy - Fast storage and search for parsed emails
• Email - Postmark Inbound Webhooks for real-time email ingestion
• 🧩 Modules:
  • Orders: Parse and confirm orders
  • Approvals: HR and budget workflows
  • Support: AI-tagged issues with urgency
  • HR Requests: Leave and policy handling
  • Enquiries: Customer request viewer
  • AI Chat Assistant: Modal chatbot for agents
• Features : Custom parsers for orders, approvals, and support tickets AI-powered accurate summary of Mail content and tag extraction Dynamic dashboards and modals Enterprise AI chat with recommended prompts

1️⃣ Inbound Email Parsing via Postmark
When a mail is sent to the Postmark Inbound Email Stream, the Postmark parses the message and sends a full JSON payload.

{
"From": "[email protected]",
"Subject": "Leave Request: July 2-5",
"TextBody": "Hi, I’d like to apply for leave from July 2 to 5 for personal reasons.",
"Attachments": [],
"Headers": [...],
...
}
It also shows if the Email follows the DKIM Signature and SPF Policy for making Email secure. I made this JSON data to be received by /inbound-email and inserted into the emails table with all metadata and raw content stored.

I wanted to distinguish Email types because of my system design. I wrote a basic logic of Keys for Order Emails ODR, HR Workflows, and Approval Emails APL, Customer Enquiries as SUP with including YYYYMMDD00serialNumber.

2️⃣ Intelligent Email Understanding with Gemini
The Email type is to be identified. I thought of using RegEx, a dictionary of known type of words and phrases. To improve the accuracy of classification, alongside parsing the JSON to use the Google Gemini LLM Model, the inbound email is parsed by a prompt chain to extract:

Type: ORDER, SUPPORT, APPROVAL, HR, ENQUIRY
Summary: A 1–2 line abstract
Tags: e.g. ["leave", "login issue"]
Criticality: Low, Medium, High, Urgent
These fields are stored alongside the email and used to guide logic and UI rendering.

3️⃣ Backend Logic & Storage (FastAPI + SQLAlchemy)
The backend routing of each email to trigger Workflows is important. The Dashboard is dynamic for real-time Email communications in an Enterprise. For Inboxops to serve scalability, FastAPI is the best choice.
The routings are as follows:
/u - The Inboxops Inbox
/dashboard or /u?page=dashboard - Support, Issues, Ticket, success rate, Dashboard Panel
/new - Customer Support Tickets Chat for the concerned team to have an in-depth chat
/support-customer - Tickets Management (Email to Ticket).
/ai-chat - Popup Enterprise AI LLM Chat that uses JSON context, summary, and tags to serve every query in detail. (Present at Header)

Enterprise Inboxops AI LLM Chat:

Example Queries

What is the latest Approval Mail about?
What do the emails discuss in the quarter 2025 - 2026?
What is the time when Employees take the most leave, and how does it affect the Customer Enquiries Supports Tickets?
Track me order ODR-202606030002

The last thing in development is continuous development, improvement.
From the moment an email lands, it’s transformed into a clean JSON payload with headers, bodies, metadata, and attachments — ready to power whatever comes next.

That kind of reliability made it easy to focus on the product, not the plumbing.
I built Inboxops on this foundation.
Here are the next features on the roadmap:
Add Slack/MS Teams integration Inbound Email to Slack MS Teams and Slack MS Teams to Outbound Email using Postmark.
Sentiment analysis for customer complaints.
CRM Integration for Internal Team Dashboards

Well, if you made it this far in the post, here is a thank you 💛 to you 😁✨