Postmark Email Agents

ConnectedNatural — Mon, 09 Jun 2025 06:57:51 +0000

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

What I Built

I built "Postmark Email Agents," a powerful email processing application that intelligently handles incoming emails to automate tasks and organize your digital life. The system receives emails through Postmark webhooks, analyzes their content using a suite of AI agents, and automatically identifies and processes actionable items.

Whether it's an event to be added to your calendar, a shipping notification for a recent purchase, or important information that needs to be saved as a note, our application handles it all. It's designed to turn a cluttered inbox into a structured, automated workflow, making sure you never miss an important task hidden within an email.

Demo

Frontend is not ready yet

Code Repository

The full source code for this project is available on GitHub:
GitHub Repository

How I Built It

The application is built with a modern, scalable architecture using Python and FastAPI. Here's a breakdown of the implementation process and technology stack.

Implementation Process

The development process was centered around a modular, decoupled architecture to ensure maintainability and extensibility.

Foundation with FastAPI: I started by setting up a FastAPI application to serve as the backbone for the API. This included configuring versioned routes, CORS middleware, and a static file server for attachments.
Email Processing Pipeline: The core of the application is the email processing pipeline. I implemented a WebhookProcessingService that listens for incoming Postmark webhooks. This service is responsible for:
- Receiving and storing the raw email JSON for reliability.
- Parsing the email content, headers, and attachments.
- Organizing emails into conversation threads.
- Saving attachments to the file system and making them accessible via a URL.
AI-Powered Actionables Module: The most innovative part of the project is the actionables module. I designed a multi-agent system using a Pydantic AI Agents framework. This is how it works:
- An Agent Orchestrator processes the content of each new email.
- It uses specialized agents, each defined by a Pydantic model, to look for specific types of information. For example, the CalendarAgent has a Pydantic model that defines fields like event_name, date, and attendees.
- The orchestrator sends the email text and the Pydantic schemas to a LLM (Gemini 2.5 Flash), using its function-calling ability to extract structured data.
- If the LLM successfully populates an agent's Pydantic model, the orchestrator triggers the corresponding action, like adding an event to a calendar. This design makes it incredibly easy to add new agents for new tasks without changing the underlying system.
Database and Data Access: I used PostgreSQL as the database and SQLAlchemy as the ORM. The repository pattern was implemented to abstract database logic from the business logic, making the code cleaner and easier to test.

Technology Stack

Backend: Python, FastAPI
Database: PostgreSQL
ORM: SQLAlchemy
Data Validation: Pydantic
AI/Agents: PyDantic AI Agents
Email Service: Postmark (for inbound email processing via webhooks)

Experience with Postmark

Postmark was a pleasure to work with. The webhook integration was straightforward, and the documentation was clear and comprehensive. The detailed JSON payload provided for each email made it easy to parse and extract all the necessary information, from headers and body content to attachments. The reliability of the webhook delivery was crucial for the success of this project.

Autonomous AI Research Analyst using Runner H

ConnectedNatural — Sat, 07 Jun 2025 06:11:17 +0000

This is a submission for the Runner H "AI Agent Prompting" Challenge

What I Built

I built an Autonomous AI Research Analyst using Runner H. This workflow completely automates the process of tracking the entire AI landscape for a given week. It systematically scans dozens of sources, identifies the most significant product and company launches, filters them based on rigorous criteria (like traction, funding, and founder reputation), and generates two comprehensive intelligence reports in Google Docs.

Finally, it schedules a recurring weekly review meeting in Google Calendar, complete with links to the fresh reports, and notifies me on Slack when the entire process is complete. It solves the critical problem of information overload and the immense manual effort required to stay ahead in the fast-paced world of AI.

Demo

Scanning in Action: The agent's browser activity log would show it methodically visiting Product Hunt, Hacker News, TechCrunch, and other key sources.
Data Compilation: A draft document would populate in real-time, listing every new AI tool, company, and open-source project it finds, along with preliminary details.
Intelligent Filtering: The agent would then be seen cross-referencing this list against its criteria. It would highlight entries with over 500 Product Hunt upvotes, a new seed round from a major VC, or those founded by a well-known AI researcher.
Automated Reporting: Two Google Docs, "AI Landscape Weekly Report" and "Promising AI Launches," would be created from scratch. You'd see the agent populate them with structured data, analyses, and direct source links.
Calendar & Slack: A new event, "AI Landscape Weekly Report Review," would appear on Google Calendar for Monday at 11:00 AM. Simultaneously, a Slack message would pop up in a designated channel, announcing that the reports are ready and providing the links.

This entire sequence runs from start to finish with zero human intervention.

How I Used Runner H

I leveraged Runner H's capabilities for web Browse, data extraction, and deep integration with Google Workspace and Slack. My prompt is structured as a precise, multi-phase operational plan that the agent executes flawlessly.

Here’s the step-by-step breakdown of the instructions given to Runner H:

Phase 1: Pre-Report Research Phase (Data Gathering & Verification)

Comprehensive Scanning: Runner H is instructed to browse a prioritized list of sources (Product Hunt, Hacker News, TechCrunch, Twitter/X, etc.) to find all new AI products, companies, and open-source projects from the past week.
Categorization & Documentation: As it finds launches, it categorizes them (Generative AI, LLMs, MLOps, etc.) and documents them in a temporary draft file, ensuring no finding is lost.
Verification: The agent cross-references information across multiple sources to verify the accuracy of launch details, funding, and founder information.

Phase 2: Selection of Promising Launches (Intelligent Filtering)

Criteria-Based Evaluation: Runner H reviews the draft list against a strict set of criteria for significance:
- High Initial Traction (>500 Product Hunt upvotes, >100 Hacker News points).
- Significant Funding (announced round with a notable lead investor).
- Prominent Founder(s) (track record in AI or successful exits).
- Novelty/Breakthrough (introduces a new technique).
- Addresses Critical Pain Point.
Selection & Profiling: It selects launches that meet at least two criteria and prepares detailed profiles for them, compiling all relevant data points for the final report.

Phase 3: Report Generation (Google Docs Integration)

Main Report Generation: Runner H creates a Google Doc titled AI Landscape Weekly Report - YYYY-MM-DD. It structures the document with a title page, executive summary, table of contents, and detailed sections for all new products, companies, funding rounds, and open-source projects. It includes an appendix of all raw URLs scanned.
Promising Launches Report: It generates a second, more focused Google Doc titled Promising AI Launches - YYYY-MM-DD. For each selected launch, it creates a detailed entry covering its summary, reasons for being promising, key differentiators, market impact, and team/funding details.

Phase 4: Calendar Integration & Notification (Automation)

Event Creation: The agent accesses Google Calendar and creates a recurring event for every Monday at 11:00 AM IST titled "AI Landscape Weekly Report Review - Week of YYYY-MM-DD".
Dynamic Description: It populates the event description with placeholder text and inserts the shareable links for the two Google Docs it just created.
Slack Notification: Upon successful creation of the documents and the calendar event, Runner H sends a message to a designated Slack channel, announcing that the reports are ready and providing the direct links.

Use Case & Impact

This workflow is a game-changer for anyone who needs to maintain a strategic pulse on the AI industry.

Who benefits?
- Venture Capitalists & Investors: Get a weekly, pre-vetted list of promising startups and trends, saving dozens of analyst hours.
- Product Managers & Strategists: Track competitors, identify emerging technologies, and discover new integration opportunities without the manual grind.
- Tech Journalists & Researchers: Receive a structured brief with verified sources, dramatically speeding up their reporting and analysis workflow.
- AI Enthusiasts & Founders: Stay on top of the latest tools, papers, and funding announcements effortlessly.
How does it improve existing processes?
- Saves 15+ Hours Weekly: Eliminates the laborious, manual task of scanning, copying, pasting, and formatting information from dozens of tabs.
- Ensures Comprehensive Coverage: The agent's systematic approach is more thorough and less prone to human error or bias than manual research.
- Delivers Actionable Intelligence: By not just listing launches but filtering for "promising" ones, it turns raw data into strategic insight.
- Creates a System of Record: The automated reports create a clean, week-over-week archive of AI industry evolution.

This agent transforms a chaotic, time-consuming research task into a reliable, automated intelligence pipeline, freeing up key personnel to focus on high-level analysis and decision-making rather than data collection.

Social Love

Full Prompt

The complete prompt provided to Runner H is below.

# Research Workflow

## 1. Pre-Report Research Phase

**Objective:** Conduct thorough research to identify and verify all relevant AI product and company launches from the past week. Ensure at least **10** unique products or launches are identified, regardless of company or launch type.

**Actions:**

- **Comprehensive Scanning:** Use prioritized sources (e.g., Product Hunt, Hacker News, Reddit, Twitter/X, TechCrunch, VentureBeat, Axios Pro, Google News, Tech Blogs, Medium/Substack, LinkedIn, Indie Hackers, GitHub) to gather information on new AI products, companies, and open-source projects. Aim for at least **10 unique products** to be included, regardless of the type or the company behind them.

- **Categorization:** Organize the findings into categories such as:
    - Generative AI
    - AI Agents
    - LLMs
    - MLOps
    - AI Infrastructure
    - AI for X
    - Computer Vision
    - NLP
    - Reinforcement Learning
    - Responsible AI
    - Ethics

- **Verification:** Cross-reference information to ensure accuracy and reliability. Ensure product validity by checking against multiple credible sources.

- **Documentation:** Maintain a draft document listing all identified launches with preliminary details, ensuring **at least 10 entries**. This will allow flexibility and variety across different types of launches.

---

## 2. Selection of Promising Launches

**Objective:** Identify and highlight launches that meet specific criteria indicating significant impact or innovation. Prioritize **quality over quantity**, but still select at least **5 products** that meet the criteria.

**Criteria:**
- **High Initial Traction:** Look for over 500 upvotes on Product Hunt within 24 hours, over 100 points on Hacker News, or widespread positive discussion on Reddit/Twitter.
- **Significant Funding:** Focus on launches with a notable funding round (e.g., Seed, Series A) from recognized investors.
- **Prominent Founder(s):** Preference for founders with a proven track record in AI or successful exits.
- **Novelty/Breakthrough:** Look for products that introduce a new approach or solve a previously intractable problem.
- **Addresses Critical Pain Point:** The launch should address a significant problem for a large market segment.
- **Strong Community Engagement:** Evaluate if there is an active community or high engagement surrounding the product.

**Actions:**

- **Review Draft Document:** Evaluate the entries in the draft document against the above criteria to ensure they meet at least two of these points, or one with substantial impact.
- **Select Promising Launches:** Choose a minimum of **5 promising launches** that meet the criteria.
- **Prepare Detailed Profiles:** For each selected launch, compile a comprehensive profile including:
    - Name of the launch
    - Product type and category
    - Summary of the launch
    - Reasons for being promising
    - Key differentiators
    - Potential market impact
    - Team and funding spotlight
    - Direct link to the launch
---

## 3. Report Generation

**Objective:** Create detailed and structured reports summarizing the findings of the past week’s AI product launches.

**Actions:**
- **Main Report:**
    - **Title Page:** Include the report title, date range, and author information.
    - **Executive Summary:** Provide a concise overview of key trends and highlights.
    - **Table of Contents:** Automatically generated for easy navigation.
    - **Sections:**
        - **Promising Launches & Major Updates:** Detailed analysis of selected promising launches.
        - **New AI Products & Features:** Chronological or domain-based listing of new products and features.
        - **New AI Companies & Funding Rounds:** Details of new companies and funding announcements.
        - **Noteworthy Open-Source Projects & Research Tools:** Information on significant open-source projects and tools.
    - **Appendix:** Include a list of raw URLs scanned and any encountered errors.

- **Promising AI Launches Report:**
    - **Title Page:** Include the report title and date range.
    - **Entries:** For each promising launch, provide a detailed analysis including:
        - Launch name
        - Type (product/company)
        - Summary
        - Reasons for being promising
        - Key differentiators
        - Potential market impact
        - Team and funding spotlight
        - Direct link to the product/company

---

## 4. Calendar Integration

**Objective:** Schedule the recurring weekly report review meeting for proper tracking and accountability.

- **Event Creation:**
    - **Title:** `AI Landscape Weekly Report Review - Week of YYYY-MM-DD`
    - **Date & Time:** Every Monday at 11:00 AM IST
    - **Duration:** 30 minutes
    - **Description:** Provide the following details in the description:
        ```


        Review the latest AI product and company launches. Access the reports via the links below:
        Main Report: [Link to 'AI Landscape Weekly Report - YYYY-MM-DD']
        Promising Launches: [Link to 'Promising AI Launches - YYYY-MM-DD']


        ```
    - **Attendees:** Add `[connectednatural@gmail.com]`
    - **Reminder:** Set a 15-minute notification before the event.

---

## 5. Operational Instructions & Error Handling

**Objective:** Ensure smooth execution and proper error handling during the process.

- **Execution Order:**
    1. Conduct comprehensive research and data extraction.
    2. Filter and categorize launches (ensure at least 10).
    3. Generate the "AI Landscape Weekly Report" Google Doc.
    4. Generate the "Promising AI Launches" Google Doc.
    5. Obtain shareable links for both documents.
    6. Create the Google Calendar event.
    7. Notify the user via Slack about report completion.

- **Error Handling & Logging:**
    - Maintain an "Error Log" for any issues such as failed URL access attempts, incomplete information extraction, or problems with document or calendar creation.
    - Implement a **retry mechanism** for unresolved issues; notify the user if the error persists.

- **Notification Protocol:**
    - **Slack Notification:** After successful completion, notify the Slack channel (#ai-insights):
        ```


Weekly AI Landscape Report for [Week of YYYY-MM-DD] is ready. Reports and Calendar event created.


        ```
    - Include direct links to both Google Docs in the Slack message.

---

## 6. Data Verification

**Objective:** Ensure that all data presented is verified, accurate, and objectively reported.

- **Source Citation:** Always include the Primary Source URL for each entry to allow for verification.
- **Objectivity:** Present information in a neutral, factual manner. Avoid sensationalism or biased language.
- **Transparency:** If information is inferred or uncertain, indicate this.
- **Data Integrity:** Prioritize information from official announcements and reputable tech news sources. Note any discrepancies or prioritize the most credible source when conflicts arise.

DEV Community: ConnectedNatural

Postmark Email Agents

What I Built

Demo

Code Repository

How I Built It

Implementation Process

Technology Stack

Experience with Postmark

Autonomous AI Research Analyst using Runner H

What I Built

Demo

How I Used Runner H

Use Case & Impact

Social Love

Full Prompt