Boris Mann’s Personal Blog

May 2017

​​Cully: We’re building a bot!

​​Cully: We’re building a bot!

​​Frontier Foundry partners with 20 Year Media, building messaging + machine learning engines

​​We’re excited to be announcing that we’re starting work on our first messaging + machine learning product at Frontier Foundry.

We’re partnering with ​​20 Year Media to build a team to tackle a consumer facing messaging bot.

​​Cully will live on multiple messaging platforms and recommend great events

We think of Cully as a cross between a knowledgeable concierge and your cool friend that has the scoop on new things to try out.
Getting recommendations on where to meet friends later

We’ll start by indexing Vancouver for evening events. All of us only have so many evening “slots” during the week or weekend to watch a movie, go to a community meetup, or see a dance performance. Some will be with coworkers, others with family, and of course with your friends. Cully will be exploring the concepts of the best events for you, but also when is best to go see them. That might mean pinging you about something tonight, or making sure you’ve got a good mix of work/family/fun booked in the next couple of weeks.

Make sure to fit in some friend hang out time

Conversational front ends

One of the big challenges with bots is their interface – their “front end” – is very basic and lacks context. Users become frustrated when they ask a question and the bot doesn’t understand, or answers one question and doesn’t understand what seems to be a logical next question.

Human: “What’s the weather in Boston?”
Bot: “It’s raining”.
Human: “Should I bring my umbrella?”
Bot: “I don’t understand what you mean.”

Natural Language Processing (NLP) and great conversational dialogue design are needed. We aren’t claiming we’re going to solve this, but we have some ideas about an opportunity for better design tools, much like front end frameworks led to richer web applications.

Machine learning’s role in bot experiences

​​Many marketers have flooded into bots thinking about it as another push / broadcast channel. But instead we need to be thinking about these interfaces as personal agents that learn relevance, timing, and an individual’s taste over time. This means learning more about the user, remembering their past (and current!) context, all leading to more relevant recommendations.

​​This is where some of the machine learning comes in. We could be saying artificial intelligence (AI) but we very much agree with Katherine Bailey on this point:

​​What is Artificial Intelligence? It’s a meme; an impressively resilient and fecund meme. No sooner does it land in the brain of one unsuspecting human than countless new #ArtificialIntelligence tokens are spawned and sent out into the world. – Hashtag Artificial Intelligence

So we’ll stay away from that hashtag meme from now, and just say that we’ll be working on better learning, feedback loops, and interactions to recommend great events for you, at the right time and place.

Come on, let’s go have some fun!

We’re hiring for three positions for the Cully team now: lead product designer, senior tech lead, and an intermediate developer. You’ll be working with the Frontier Foundry team out of our space in downtown Vancouver, building out messaging, machine learning, and conversational front ends.

An early prototype of Cully with a basic web interface is live now and filled with Vancouver events. Check it out at, feed us some event recommendations!