# The Copilot for airline operations

Cancel flights, refund bookings, create reservations from a passport photo - in plain language. Live across every airline running on Farel, in five languages.

- Operates flights and bookings
- Creates bookings from passport scans
- Confirms before every action
- Five languages

## Most airline systems were designed to be clicked through
Every refund is a screen. Every cancellation is a workflow. Every new hire is months of training on menus.

### Click-heavy workflows

Refunds, cancellations, and changes each take minutes through legacy PSS UIs. At call center scale, that's the whole budget.

### Months of training per agent

Dozens of screens, conditional flows, edge cases. Every hire restarts the curve. Every turnover resets it.

### English-only interfaces

Most PSS UIs ship in one language. Your call center, airport staff, and ops control don't all work in it.

### IRROPS handled one-by-one

A delay or cancellation hits hundreds of bookings. Agents process them one at a time, in sequence, while passengers wait.

## One Copilot. Every operation. Every role.

Farel Copilot operates on the same data as the rest of the platform. Type what you want done - refund a booking, cancel a flight series, create a reservation from a passport photo - and the system shows you the impact, then commits.

## What Copilot does today

Operational scope, in one place. Same data and permissions as the rest of Farel.

- Flight operations - Cancel, reschedule, open or close sale, edit pax quota, edit departure time. One flight or every flight in a period.
- Booking operations - Create, refund pax-segments, edit remarks, process involuntary changes. One booking at a time or many at once.
- Booking from a passport photo - Upload an ID, get a full reservation in 20-30 sec. Replaces 2-3 minutes of manual field entry.
- Bulk operations in one command - Refund fifty bookings, cancel a flight series, requote a season. One command does what a batch job or a queue of agents would handle elsewhere.
- Plain-language targeting - "Next Tuesday," "this whole series," "all LAX flights this week." No filters, no query syntax. Copilot resolves the target and shows what it'll touch before it acts.
- Five languages, more coming - English, Russian, Spanish, Portuguese, French. Agents type in their language; Copilot replies in the same one.

## Cancel a flight, every booking follows

One command can target a flight series, a date range, or a single departure. Booking-side impact - involuntary operations, rebooking, notifications, remarks - flows automatically through the rest of the platform.

## From passport photo<br/>to issued ticket in 20 sec

Upload a scan; Copilot extracts name, DOB, document number, nationality, expiration. Agent confirms the flight and price; ticket issues. 2-3 minutes of typing collapsed into seconds.

## Nothing executes until you confirm

Refunds show the exact amount per passenger and segment. Cancellations show every booking affected. Quota changes show the new inventory state. The system commits only after the agent confirms, and stamps the action with their user identity. "Cancelled by [name] via Assistant" appears on the operation log, exportable for audit.

## Holds context across<br/>the session

Look up a booking, refund it, then start on the next one - Copilot remembers what you were just working on. Multi-step flows don't restart at every command. At Asman, 168 sessions in a month completed end-to-end without falling back to the interface.

## Already running at scale

Live across every airline on Farel. Used daily by call center, airport, and ops teams.

## Before you decide
Answers to the questions we hear most from airline teams

### How is this different from a chatbot bolted onto a legacy PSS?

Copilot operates on the same records as inventory, reservations, DCS, and payments. Legacy PSS chatbots query through APIs and middleware, which means they're limited to what those APIs expose and lag the underlying data. Farel Copilot has the same access as the interface - because it's the same platform.

### What happens if it misinterprets a command?

Every destructive action - refund, cancel, IRROPS - shows the full diff before execution. The agent sees exactly what will change and confirms before commit. If the interpretation is wrong, the agent rejects it and rephrases.

### Who can use it, and what can they do?

Same permission model as the rest of Farel. A call center agent has call center permissions in Copilot. An ops controller has ops permissions. Same access rules as the UI.

### How is this audited?

Every operation logs user identity, timestamp, source, and before and after state. "Cancelled by [name] via Assistant" appears on the operation history. Exportable for internal review or regulator request.

### Which languages does it support?

English, Russian, Spanish, Portuguese, French. Adding more on customer request.

### What operations aren't supported yet?

Coverage is published and updated as scope grows. Today's gaps include flight creation, pricing edits at the flight level, and crew assignment - these stay in the schedule and flight interfaces.