BotXperts asked — with Luis Novo of Skore
This is the fourth interview of a series of interviews I’m doing with 10 experts of the worldwide bot scene. These interviews are part of my research which aims to find guidelines for designing messenger bots. If you want to read more about what I do and why you can do it in my previous article. You can also read the previous interviews with:
Luis Novo & Skore
Luis is the CEO of Skore, a SaaS company, which aims to improve knowledge flow with their products.
We enable teams to organise the knowledge available to them, understand how information influences each member as well as the group, and strategically improve their capacity to generate better results. In a nutshell, we bridge the gap between fresh, quality content and those who need it.
About a year ago they started building Paperbot because lot’s of knowledge which was shared in their Slack team before got lost after some time. Thats why Paperbot organises the most important links shared in Slack channels and makes them available anytime. Additionally it sends a daily or weekly summary into a web interface which can be viewed from everywhere.
Right now Paperbot is used in 85 different countries which includes the US, Brazil, Germany, UK and France and impacts several thousands of users every month. After their launch in March 2016 it took them a year to reach their first 1,000 active teams and grow snow rapidly. Paperbot is built with Node.js and a Ruby/Erlang backend while the mobile apps use React Native. The Search itself is powered by Algolia and the database is exposed through GraphQL.
Bot Development
Which problems did you face while developing bots?
Handling scale has become important very early on. Because we handle a huge amount of connections with our customer and, as Slack users, they expect it all to work as fast as messaging.
What kind of problems do you see in the interaction between people and bots?
Building the right kind of interaction between users and bots is quite challenging. People expect a bot to be ‘conversational’ at times and prefer for it to be extremely practical (think buttons) at others. Knowing when to use one style versus the other requires a lot of A/B testing. That’s why we’ve implemented some pretty powerful data analytics and A/B testing tools.
What have you improved on your bots since the release?
At first, Paperbot was basically built to collect links from Slack and send email digests. Over time, we realized that it could do much more for the user. Now, it automatically categorizes each link into a searchable web interface and supports additional tags to provide more context. We’ve also released a mobile app for iOS and Android to improve user experience on-the-go. In addition, Paperbot can now make recommendations to amplify the reach of the most relevant content available on the team’s workspace. We’re working hard to make our AI-based recommendations smarter so we’re all very excited about the new releases coming up soon.
The Future of Bots
How do you see the future of bots?
BOTs aren’t the silver bullet and we have this tendency to push a lot of expectations to every new technology that comes out and then become frustrated when it fails to meet that expectation. If you look at the unique value a bot can bring, two come to my mind immediately.
- the conversational aspect: the whole idea of having ‘conversations’ with bots is powerful in the sense that you don’t need to figure out complex user interfaces and learn to make advanced queries and interactions, you can talk your way through your needs and have them fulfilled.
- we relate to BOTs as ‘entities’ in a sense that we have a natural tendency to try to talk to them and treat them as people. This means that, if you gain your users’ trust, they might provide more information and access to a bot than they would to an app or your company’s Facebook account.
So my take is that the bots of the future will earn our trust and gain access to our most private information (think health, finance, immigration). This can make them extremely powerful in helping us solve issues that other apps couldn’t so far. And by using conversation as a much more ‘accessible’ interface, they’ll bring this value to a lot of people, even those without much experience using tech.
What has to be improved so bots can work better?
Natural Language Processing is an obvious topic that a lot of companies are spending a lot of resources on improving. This is making solid steps forward and will continue to do so. Another challenge is discovery. With these many BOTs available, how will users discover the right one to help them at any given moment? Facebook and Slack have both mentioned this as priorities for them.
Bots will replace websites
What do you think about this statement?
Over-rated. BOTs are not the silver bullet and are not the right solution for everything. I think we need to look that, in the future, bots, apps and websites won’t be that different from each other, but not that one will replace the other.
Follow Luis on Twitter so you don’t miss the latest trends in messaging:
