Trainline sets out a personalised, proactive experience

Handles 350 travel searches a second

Trainline’s director of machine learning and AI, Reda Kechouri, outlined in Paris on 3 June how the London-based company is using artificial intelligence across the user experience, distribution and internal productivity. The aim, he said, is to become more than an app where people buy tickets and instead act as an intelligent travel companion built on three strategic pillars.

The company is rolling out tools to give travellers greater visibility and real-time information. Its Travel Forecast solution sends notifications to users and, when problems arise, offers alternatives. The system is currently operating in England and, once it is mature, is expected to be deployed in other countries.

“Our model can predict the state of the rail network two hours in advance,” said Kechouri. “If there is an incident somewhere, we know it will have a knock-on effect and we are able to provide a better prediction. All notifications are sent by this same system and we use LLMs to make them customisable and relevant for the user.”

Kechouri also discussed Trainline Assistant’s ability to automate online support as far as possible. Nearly one third of refund requests (32%) are processed automatically, he said. “People use this feature to ask very simple questions they would never have phoned about,” Kechouri noted. “I don’t think anyone is going to call customer service to ask whether there are toilets at Gare du Nord, for example.” Automation does not simply replace human assistance; it can also complement it.

350 searches a second

Trainline is investing heavily to be present wherever users search, including in emerging AI systems such as ChatGPT, Claude and Gemini. The company offers an app within ChatGPT that lets users plan a complex itinerary in natural language, with the AI pulling options from Trainline to complete the purchase. This distribution strategy extends to so‑called “Search AI”, where the company works with platforms such as Reddit to ensure its solutions surface as the most relevant when travellers make queries.

The final pillar, dubbed “acceleration”, is changing how Trainline creates value. Generative AI is used by marketing teams to reduce the cost of producing content and images. “The cost of code is falling,” Kechouri said, which allows faster testing of new ideas, including voice integration with ElevenLabs.

This multi‑level adoption of AI is enabled by Trainline’s scale. The platform has 27 million active users across 40 countries and connections to 270 operators. “Every second Trainline processes 350 travel searches,” Kechouri said. Each month, around 3.8 million journeys are analysed, creating a virtuous circle: the more data there is, the more accurate the AI models become in delivering the best prices and a smooth experience. That critical mass, Kechouri added, helps Trainline tackle complex challenges while building in ethical and safety safeguards so that AI remains a reliable aid and never harms the user.