How to Create AI Labels in Orqon: A Complete Guide to Smart Email Categorization

What Are AI Labels?

AI Labels are the core feature of Orqon. Unlike traditional email filters that rely on static rules like "if sender contains X, move to folder Y," AI Labels use natural language understanding to analyze the context, intent, and meaning of every incoming email — and automatically assign the right category.

Think of it this way: instead of writing complex filter rules, you simply describe what kind of emails you want to catch, and Orqon's AI does the rest.

Getting Started: Navigating to Label Management

1. Log in to your Orqon dashboard
2. Click Mail Portal from the sidebar
3. Navigate to the Labels tab
4. Click the "Create Label" button

You'll see a form with two main fields: Label Name and Label Prompt.

Understanding Label Names

The Label Name is the visible tag that will appear on your emails. Keep it short, clear, and descriptive:

• ✅ Shopping — Clear and concise
• ✅ Work/Urgent — Categorized with priority
• ✅ Finance & Billing — Descriptive grouping
• ❌ Emails I need to read later maybe — Too long and vague

Mastering Label Prompts

The Label Prompt is where the magic happens. This is your instruction to the AI — a natural language description of what kind of emails should receive this label. The better your prompt, the more accurate your labeling will be.

Basic Prompt Structure

A good label prompt answers the question: "What kind of emails should get this label?"

Here's the basic format:

Emails that [describe the content, sender type, or purpose]

Prompt Examples by Category

📧 Shopping & Orders

Label Name: Shopping

Prompt:

Emails related to online shopping, order confirmations, shipping notifications, delivery updates, purchase receipts, and return or refund communications from e-commerce platforms and retailers.

💰 Finance & Billing

Label Name: Finance

Prompt:

Emails about financial transactions, bank statements, credit card alerts, invoice payments, subscription billing, tax documents, and any monetary notifications from financial institutions or payment services.

🔔 Social & Notifications

Label Name: Social

Prompt:

Notifications from social media platforms like LinkedIn, Twitter, Instagram, and Facebook. Includes connection requests, mentions, likes, comments, direct messages, and follower updates.

💼 Work & Professional

Label Name: Work

Prompt:

Professional and work-related emails including meeting invitations, project updates, task assignments, team communications, client correspondence, and internal company announcements.

📰 Newsletters & Marketing

Label Name: Newsletter

Prompt:

Marketing emails, newsletters, promotional offers, product announcements, blog digests, weekly roundups, and subscription-based content from companies and publications.

🚨 Security & Alerts

Label Name: Security

Prompt:

Security-related emails including login alerts, two-factor authentication codes, password reset requests, suspicious activity warnings, account verification emails, and security breach notifications.

Advanced Prompt Techniques

Be Specific, Not Generic

The more specific your prompt, the better the AI performs:

• ❌ Generic: Important emails
• ✅ Specific: Emails from clients requesting urgent changes to active projects, or emails containing deadlines within the next 48 hours

Use Exclusion Language

You can tell the AI what not to include:

Emails related to software development including pull requests, code reviews, CI/CD notifications, and deployment alerts. Exclude marketing emails from developer tool companies.

Combine Multiple Signals

Describe both content and context for maximum accuracy:

Emails that contain travel itineraries, flight confirmations, hotel reservations, car rental bookings, or visa-related documents. These typically come from airlines, booking platforms like Booking.com or Airbnb, or travel agencies.

Language-Aware Prompts

Orqon's AI understands multiple languages. You can write prompts in your preferred language or specify language-based filtering:

Turkish language emails related to government services (e-Devlet), tax notifications, SGK updates, and official correspondence from Turkish institutions.

Real-World Label Setup: A Complete Example

Here's how a typical professional might set up their labels:

Tips for Better Label Accuracy

1. Start broad, then refine — Create a general label first, observe what it catches, and fine-tune the prompt based on misclassifications

1. Review regularly — Check your labeled emails weekly to spot any patterns the AI is missing or incorrectly categorizing

1. Avoid overlapping prompts — If two labels have very similar prompts, emails might get inconsistent labeling. Make each prompt distinct

1. Use real examples in your mind — When writing a prompt, think of 5-10 actual emails you've received that should match. Does your prompt description cover all of them?

1. Keep prompts updated — As your email patterns change (new projects, new subscriptions), update your label prompts accordingly

What Happens After Creating a Label?

Once you create a label:

1. Immediate activation — The label starts working on all new incoming emails instantly
2. AI analysis — Each incoming email is analyzed against all your label prompts
3. Automatic tagging — Matching emails receive the label tag automatically
4. Dashboard visibility — Labeled emails appear in your Mail Portal with their assigned tags
5. Filter & search — Use labels to filter your inbox and find emails instantly

Conclusion

Creating effective AI labels in Orqon is all about writing clear, descriptive prompts that tell the AI exactly what you're looking for. Start with the examples above, customize them for your workflow, and iterate as you discover what works best for your inbox. The more precise your prompts, the more time you'll save — and the closer you'll get to a truly automated email experience.