📬 inbox-to-rag

Convert your XML email archive into a smart RAG (Retrieval-Augmented Generation) source using Node.js + OpenAI

---

🔍 What is this?

inbox-to-rag is a lightweight Node.js script that takes an XML file containing your emails and lets you ask smart questions about them — powered by OpenAI embeddings + GPT.

Great for:

• Searching your inbox intelligently
• Prototyping RAG workflows
• Working with email archives

---

💡 How it works

1. 📥 Load and parse emails.xml
2. 🧠 Create vector embeddings with OpenAI
3. 🧾 Compare your question to the email content
4. 🤖 Feed the most relevant chunks into GPT-4 for the final answer

---

⚙️ Setup

1. Clone this repo

git clone https://github.com/rastmob/inbox-to-rag.git
cd inbox-to-rag

2. Install dependencies

npm install

3. Create your .env file

OPENAI_API_KEY=your-openai-key-here

You can use .env.example as a starting point.

4. Add your emails.xml

Drop your XML file in the root of the project. The structure should look something like:

<emails>
  <email>
    <subject>Invoice Reminder</subject>
    <from>john@example.com</from>
    <date>2024-12-01</date>
    <text>Just a reminder about the unpaid invoice from November.</text>
  </email>
  ...
</emails>

---

🚀 Run it

npm start

You can change the sample question in index.js:

const answer = await askQuestion('What did John say about the invoice?', vectors);

---

🧠 Example output

🧠 Answer:
John reminded you about an unpaid invoice from November. He likely expects a follow-up or payment soon.

---

📦 Technologies Used

• Node.js
• OpenAI Embeddings + GPT-4
• fast-xml-parser
• dotenv

---

✨ Author

Made with ❤️ by Mehmet Alp
📬 For software development related emails: mehmet.alp@rastmobile.com
🔗 Rast Mobile
👨‍💻 GitHub: rastmob

---

🪪 License

MIT — use freely, improve freely 🚀