InMotion uses cookies to store information on your computer to improve our website. One of the cookies we use is essential for parts of the site to work and has already been set. To find out more about the cookies we use and how to manage or delete them, please refer to our Privacy, Legals & Cookies page. By continuing to use the site you agree to our use of cookies.

Building chatbots for a smoother journey

Category: Technology
Published: 26/05/2017

Navigating a city as a tourist can be a frustrating experience. You’re struggling with suitcases, maybe with children in tow, and you find yourself stopping every few minutes to get your smartphone out to check you’re heading in the right direction for the bus to take you to your hotel, bumping into commuters as you do so.

Apps have made it easier to access real-time transport information but they still require quite a bit of button pressing and fumbling about. Chatbots, on the other hand, could make the task more efficient.

Get chatting

The chatbot sales pitch is that if you have a query, in the time that it takes you to download and launch the relevant app – or navigate the mobile website – you can get a reliable answer from a bot by feeding it a few keywords.

“You can Google or search for everything about anything, but you have to pull in all the information yourself,” says Lex Oudijk, founder and creator of Lewis, a recently launched virtual concierge designed to help people visiting London.

“Some people are just too busy or stressed and don’t have the time to do this. A big advantage a bot has over other platforms is that it can push information.”

Chatbots work by scraping data from websites using APIs. In the case of Lewis, the bot can give users directions to desired locations; it can also suggest best times to travel or visit attractions, using data from TfL and MasterCard.

Oudijk says that the bot’s potential to recommend the best times to visit attractions and advise when not to go can also alleviate some of the pressure on transport systems and “turn ordinary cities into smart cities”.

The global appeal

Although chatbots can be built from scratch, the majority, including Lewis, are integrated into existing text-based apps, such as Facebook Messenger, Telegram and WhatsApp.

The more a bot learns the better it should become at understanding.

Smartphone users are loyal to messaging apps and their popularity means that companies that invest in chatbots have an opportunity to reach residents in markets previously not considered.

According to research, 27% of the global population will be using a messaging app by the end of 2019, and the UK currently accounts for nearly a fifth of all messaging app users in Western Europe. Proliferation is also extremely high across Latin America. In Brazil, for example, 94% of smartphone users have WhatsApp installed on their device, and in Mexico it’s 92%.

PoolMyRide, a Delhi-based carpooling service that connects people who are planning similar journeys, has users in several dozen countries, including the US, Canada, Costa Rica, Pakistan, Malaysia and Germany.

It’s primarily app-based, but its Facebook Messenger bot can be used to search for carpools and receive both messages from other riders, as well as notifications of any future carpools.

Chatbots work by scraping data from websites using APIs.

Co-founder Abhishek Talwar says that the service is particularly useful for those living in rural areas who experience patchy 3G or 4G signal or have no WiFi access and don’t want to be charged for downloading an app that is tens of megabytes in size.

The downside

Despite the global appeal of having chatbots built into existing platforms, there are still frustrations that need to be ironed out.

“Developers don’t get as much freedom,” says Talwar. For instance, we can only add in a limited number of buttons or menu options. Coming from an app background, this is a major shift in deciding on user experience design.”

He adds that it means drilling down further to understand what users need. If they struggle to interact with a bot then they won’t use it, so it’s essential that attention is paid to language. It should be concise and conversational.

Video tutorials could be embedded so that they pop up at the start of a chat, however this could be counterproductive as it might inconvenience users, especially those who are using their mobile data.

Another frustration is that chatbots can sometimes lack intelligence. They can be built to be chatty, but they won’t necessarily understand every question or command. This means that users may have to experiment with their wording to get the desired response, which could test their patience.

Of course, over time, the more a bot learns the better it should become at understanding complicated and longer sentences, resulting in an improved service.

Your transport buddy

Ultimately, a chatbot’s machine learning element and ability to process language means that it should also be able to push more relevant content to users, says Oudijk.

A big advantage a bot has over other platforms is that it can push information.

“A chatbot lives among your friends on Facebook Messenger or your contacts on WhatsApp. And it should act like one too,” he adds.

“A bot should be that one friend you have who knows a city inside out; who’s a local who can tell you how you can get about without hassle and if there are problems with public transport, who tells you want you want to hear.”

Chatbots won’t make apps obsolete, but they have the potential to play a key role in eliminating those pesky transport-related headaches.