Making sense of chatbots

‘Chatbots’ have popped up all over the internet this year. Applications have varied from customer service to a chatbot lawyer that offers to resolve parking disputes. Victor Trokoudes, co-founder of Plum, spoke to Courier about his savings platform that deploys its particular mascot, the Plumbot.  

16 Dec 2016

How did Plum get started?

Alex [Plum’s co-founder] was downloading his bank statements and calculating a budget. We saw the transaction data was rich information, and thought, how can we make it as effortless as possible to save? The objective of the algorithm is to save very small amounts that people don’t notice, so they get into the habit of seeing money accumulate.

Why did you create Plumbot?

Chatbots are great for certain products, and ours happened to be one of those. Plum is a background process that looks at your account and saves money for you; you don’t have to do anything. Chatbot is perfect because we can communicate when you save money, and you can enquire about how much you’ve saved so far, or to proactively save money. Those are the operations we need [the chatbot] to do.

You’re asking for people’s bank details – how do you get them to trust you?

It’s definitely the biggest challenge. We mention Transferwise [where Victor worked previously] on our homepage to give people the trust that we’ve built things around money before. We’ve had a BBC article about us, Tech Crunch, Sunday Times. These things build trust.

It’s sad that people have been made to feel uncomfortable sharing these details. As a result, you can’t get certain advice you would be getting otherwise.

Why should a company use a chatbot?

If you already have an app on your homescreen, like Facebook, that you’re using every day, it could be giving you useful practical information.

In sectors where you have access to data which is frequently updated, you can draw interesting conclusions or give advice to people based on that data. Seeing a dashboard of your spend on electricity, for instance, is not that interesting. It’s more interesting to highlight specific, actionable information.

What are the biggest challenges with developing chatbots?

Chatbot is a word people don’t understand, but if you say, ‘if you want to save money, go to Plum and talk to it,’ it’s self explanatory.

People [also need to] become aware of how they should use chatbots. Right now, people presume a chatbot can understand everything, which is not the current state. As people understand more and more how to interact with a chatbot they’ll get the most value without being endlessly frustrated that the chatbot doesn’t understand them.

We’ve noticed that people who discover us online and want to use Plum to save money don’t try to test the bot. But when we tell people about the app, they try to push it, to chat to it as much as they can.