📬 From Inbox to Impact: An AI-Powered Wall of Love

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

What I Built

For this challenge, I built Wall of Love — an AI-powered testimonial showcase system that automatically receives customer feedback through email, analyzes it using AI, and displays curated testimonials on a public-facing website.

In simple terms:

👉 You send an email.

🤖 AI checks if it’s genuine.

🌟 If yes, it appears live on the site instantly.

This project is built on Postmark’s Inbound Email Streaming and showcases how easy and powerful it is to build automated experiences from something as old-school as email.

I wanted to take the traditional "wall of love" page to the next level — make it smart, real-time, and effortless to manage.

---

Demo

🔗 Live App

💌 Send a test testimonial to: [e-mail]

📸 Here’s how it looks:

You email → Postmark catches it → AI filters it → Supabase stores it → React displays it

---

Code Repository

💻 GitHub: AnanduApillAi/postmark-challenge

---

⚙️ Tech Stack

• Next.js 14 (App Router + Edge Functions)
• Postmark Inbound Webhooks (email parsing + real-time delivery)
• OpenAI GPT-4o-mini (AI filtering + sentiment analysis)
• Supabase (database + row-level security)
• Tailwind CSS (UI)
• Vercel (hosting)

---

🧠 The Smart Workflow

Here’s what happens under the hood:

1. Email Reception: Customer sends an email to [e-mail].
2. Postmark Inbound: Instantly receives it and calls my webhook with full data.
3. AI Processing:
   • Removes greetings, signatures, etc.
   • Scores sentiment (-100 to +100)
   • Filters spam and low-confidence entries
4. Storage: Only high-quality testimonials are stored in Supabase.
5. Live Display: A React-based frontend shows approved content in real-time.

---

🤖 Why Use AI?

Because not every email is a testimonial.

Some people say “hi”, others try to sell you things. AI helps clean the clutter — it only shows what truly matters: heartfelt feedback.

I’ve added confidence scoring, sentiment filtering, spam detection, and manual review flags. And yes — GPT-4o-mini is doing all of that in milliseconds.

---

💡 Why Postmark?

Before this project, I had never used Postmark. After this project, I’m not going back.

• Setup was seamless.
• Delivery is instant.
• The inbound webhook UI is intuitive.
• Email parsing "just works" — no MIME headaches, no flaky payloads.

Honestly, Postmark feels like email done right. I wish I’d discovered it earlier.

---

✨ What I Loved

• Watching a testimonial email show up instantly on the frontend. That’s a dopamine hit.
• Building an AI filter that works. Not perfect, but pretty reliable.
• Combining old (email) and new (LLMs) to create something delightfully automatic.

---

🧠 Learnings & Takeaways

• Postmark inbound + webhooks = endless automation potential.
• GPT is amazing at extracting meaning, but you still need guardrails.
• A little design polish.

---

💭 Final Thoughts

Sometimes, the best UX is no UX — just let users send a plain email, and you handle the rest. That's the magic I tried to build here.

If you're reading this, send me a note at [e-mail]. Let’s see how smart the system really is :)

Thanks to the Postmark team and the dev.to community for this challenge — I had a blast building this.

---

🔗 Links

• 🌐 Live Demo: wall-of-love.anandu.dev
• 💻 GitHub Repo: github.com/AnanduApillAi/postmark-challenge
• 📖 Postmark Docs: Inbound Email Webhooks

---

Built with ❤️ by @AnanduApillAi

Email to testimonial. Just like magic. ✨