Introducing the Seat61 chatbot, powered by Byway
We’re delighted to announce the launch of an exciting new chatbot collaboration between Byway and Seat61. After a long-standing partnership between Mark Smith (author of Seat61.com and overland travel expert) and Byway, we asked Mark what would help him and help people turn his insights into trips. Together, we came up with the idea of taking his reams of superb travel walk-throughs and decades of experience and pairing this with Byway’s routing expertise technology to make a chatbot that helps you plan trips.
The chatbot has been designed to relieve Mark of the need to respond to frequently asked questions that are already answered on Seat61’s website. For instance, if someone asks about the quickest route to Morocco, the chatbot can provide an instant, reliable response, including a link to the relevant page with all the detail Mark has already shared on Seat61.com.
Smarter than other AI chatbots for travel planning
The chatbot is built on an extensive and rigorously maintained dataset, combining Byway’s slow travel expertise with the detailed information available on Seat61.com. To ensure reliability, the chatbot is regularly updated to reflect the latest route changes, timetables, and travel advice.
Unlike general AI tools, this chatbot does not invent or hallucinate answers. Instead, it strictly retrieves information from Byway’s and Seat61’s datasets. If the answer to a question isn’t found in this trusted knowledge base, the chatbot won’t guess—it will simply direct users to where they can find more information.
Not only does the chatbot surface detailed travel information, but it also integrates with Byway’s booking technology, allowing users to build and book rail trips directly in the bot. This dual functionality means that travellers can move from researching a route to planning and booking their trip seamlessly, with Byway providing a protected package including curated accommodation (lots of which Mark also recommends), and taking care of all ticket-booking and stay-booking hassle.
Mark earns affiliate commission for trips booked through the chatbot when users follow its recommendations, making this a great partnership that helps people plan while rewarding Mark for all the fantastic work he does producing this free information.
“I’ve been a long-time advisor to the Byway team and an advocate for their work, and it’s been really interesting to collaborate on this project. We share the mission to encourage more people to opt for flight-free slow travel, but I’m but one man trying to answer everyone's questions! This bot makes it so much easier to plan your European slow travel trip and hopefully means everyone who’d normally email me with website questions can get a much faster reply!” - Mark Smith
How it was built
Using funding from an Innovate UK grant (government investment in game-changing R&D tech), Byway developed the core technology behind the chatbot, incorporating cutting-edge language processing capabilities via the ChatGPT API. However, unlike generic AI models, this chatbot is customised to work exclusively with Byway’s and Seat61’s datasets, ensuring accuracy.
“Seat61 is a beloved resource for train travel enthusiasts; knowing how many people rely on it as their go-to guide made the project exciting and meaningful. The technical challenge of ensuring the chatbot only pulled relevant data from Seat61 and Byway, preventing it from ‘hallucinating’ made this really uniquely interesting. We had to fine-tune the system to integrate the data while keeping responses accurate and reliable. It showcased how AI can sift through large datasets to deliver precise, useful answers” - Serafin Lichtenhahn, Byway Developer
AI and sustainability
Many travellers who choose rail are motivated by sustainability, and we’re similarly committed to reducing environmental impact. We’re a tech company on a mission to make flight-free holidays mainstream, and we use technology - now including large language models - to help lots of people quickly find, book and enjoy wonderful journey-based experiences.
Training, operating and using Large Language Models (LLMs) like ChatGPT requires computational resources, which has sparked conversations about their environmental footprint. However, their impact (55kg of CO2 per 50,000 queries) pales towards insignificance compared to swapping one short-haul flight for a train (175kg of CO2), meaning the saving from not flying one way is about the same as that from avoiding 160,000 ChatGPT queries. By promoting low-carbon travel and making rail travel easier to book, the chatbot helps support a shift away from higher-carbon transport modes like flying - and we’ve built it as efficiently as possible, including using already-trained LLM instead of training our own and data caching, to minimise the environmental intensity of our LLM use. We’re committed to sustainability across all aspects of our business, and this chatbot is no exception. By combining technology with a focus on efficiency and purpose, we aim to support the growing community of eco-conscious travellers.