R/GA and Ad Council, in collaboration with Reply.ai, have launched a Messenger bot that helps US citizens register to vote. And it has gotten very good traction!
During the last couple of weeks, we’ve had most of our distributed team in SF. We’ve had a lot of fun together, and at the same time got a lot of work done :)
The White House has not only created a bot, but they have open sourced the code, and said that it’s ‘an important part of furthering our mission to “meet the public where they are”’. The bot is very simple: it just asks the user to input a message for the president, and then asks for the user’s address and phone number. It’s written in Drupal, a PHP-based CMS. Not the most comfortable technology stack for a bot, but this is exciting news for the ecosystem.
With over 200M Active Users and a recent IPO under its belt, Southeast Asian Messaging powerhouse, LINE, has just seriously upped it’s Chatbot game. Sure it launched its bot platform a week before Facebook, but with new features like visual message types (very similar to Messenger’s generic carousels), an SDK, and the ability to let external services send user notifications (called Line Notify), Line is all in for Bots. They even have a bot competition with a $98K prize! We are actually migrating our current LINE integration to the new Messaging API this week.
Set of best practices and principles that Etienne Merineau (great guy, by the way) shares with this team while building bots. The most important point of all for me is: “Don’t trick users into thinking that the bot is a real person”. Quite a bunch of people in the industry do not agree with this, but I have tried to be an advocate of not tricking users. Your brand tone and personality should be there, thought.
A bot that promises to “prepare you for real-life conversations — minus the awkwardness and anxiety.” A simple use case with a huge benefit, I’d say. While the bot promises A.I., seems they are keeping the NLP to a minimum with more structured, quiz-like conversations vs. truly open-ended (smart for an inaugural use case). Well-known companies starting to make their move to bots, good.
Analysis of Marsbot: it suggest you places based on your Foursquare behavior. I actually enjoyed reading the (long) article, with examples were context, simplicity and pushing content at the right time can make a difference.
Curated with :love: from SF (till November) by Omar Pera (@ompemi) and the Reply.ai team. If you are SF, ping me for coffee. If you are outside, ping me to chat ;)