Interview with Tasytt about new Obie Slackbot
Check out the interview below with Tasytt (@tasytt) about their new Slackbot product Obie.
What is Obie?
Obie is a Slack bot. He’s kinda like Siri, but for business. He helps teams store and organize their collective knowledge.
What kind of features does Obie have?
His functionality can be broken into ‘push’ and ‘pull’. Users can pull information by asking questions and Obie accesses knowledge recorded in Slack, in your Google Drive, and on our web platform to answer them. Conversely, he can push knowledge using Flows, which are like task lists on steroids. Everything happens in a Slack DM. As an administrator, you can create a task list or lesson, then assign it to anyone (or everyone) on your team to complete.
Why did you decide to develop Obie?
Our mission has always been to help companies organize their knowledge and make it universally accessible and useful. We found that companies want to solve their knowledge problem, but they balk at introducing yet another software. They have countless point solutions for business problems and their knowledge is already fragmented into different platforms and cloud storage apps. They are quite simply tired of new software. Beyond that, most employees hate their LMS (Learning Management Systems) and the administrators hate managing them. And for good reason—they’re just an online binder with cleaner design.
So we knew that we needed to deliver information and the value you might find in learning software without more software. Users need to access the knowledge they need right away and without switching apps (and mindsets). Given all that, a Slack integration made perfect sense for us. It’s the default place where people spend their work days and they’re really working to make developer’s lives easy.
Who is using Obie so far, and for what purposes? How has the response been to Obie?
We’ve had a great response so far, we were featured on Product Hunt and in Venture Beat and have been in talks with some larger companies who want to try it (although I can’t say whom). It’s the best feeling when someone just ‘gets’ it and loves your product.
Teams have been using it to plan for events, to bridge the gap on remote teams, to deliver lessons or refreshers, for checklists or to onboard new hires. Most people really like that Obie can search their cloud drive in addition to knowledge they ‘feed’ him—that way they don’t need to manually move all their content. Our most interesting use case so far has been wedding event planning, that’s something we never would have predicted, but in hindsight makes perfect sense!
What tools did you use to build Obie?
Our original platform was built with C# and ASP.NET MVC, but for the bot, we used the MEAN stack. We’re incredibly lucky that there’s a community of early bot developers and one team (Howdy) thought to create a framework for bot development as they went along, called Botkit. Botkit is written in Node.js.
What was the biggest technical challenges with developing Obie?
Having a framework and community is great, but we were faced with a unique challenge: unlike a team building a bot from the ground up, we were connecting it to our already robust Tasytt application. We don’t want to diminish other bots and the developers behind them, but it’s one thing to build a bot that connects to a weather API or even one that connects to a database. It is quite different to build a bot, make it conversational, and connect it to both a database and web platform.
Our CTO Alex Sopinka had also never written a line of NodeJS, but since he’s a rockstar, he was able to build the majority of Obie in under a month. His protege Carolyn Chong, along with the rest of the team, did an amazing job testing and breaking the early Obie so that we could tighten him up and ensure he’s stable.
How can the local community help Tasytt make Obie a success?
If you use Slack, try him out and let us know what you like, hate or want to see! The majority of the functionality is free though we do have some premium features like reporting and analytics. It’s also a ridiculously simple setup process.
There’s been talk of chat being the “universal interface”. What are your thoughts on that? Do you see chat taking over more interfaces going forward?
There’s no doubt we’re going to see more bots and more companies choosing chat over a GUI, but we’re still very early on. Just like the first apps weren’t that useful, many of the bots out right now are pretty frivolous.
There are a few ways you can categorize bots: push vs pull; use of AI, ML, and NLP; standalone bot or counterpart to an existing app. There isn’t one combination that’s better, it depends what they’re trying to accomplish. A chat interface without NLP might work as a Calendar extension, but not for customer service. Unfortunately for bot developers, while being lean and working with an MVP is important, bots trying to ‘fake’ AI usually fail miserably.
Do you have any advice for startups in Hamilton?
Don’t let not being in the Valley, or even Waterloo, stop you from chasing your dreams. It doesn’t have to hurt your chances, but as soon as you modify your goals or let that become an excuse, it will.
Where are you hoping to take Obie next?
Right now we’re working on making it as easy as humanly possible to add information to a company knowledge base and access it right in Slack. Recording shorter form text is fairly simple, but we’re always cognizant of making it easy to attach rich media since most people find it easiest to learn from a combination of images, audio and text. Each piece of knowledge or step in a Flow needs to be short enough that people actually consume the information, but not so short that the medium becomes useless.
We recently implemented an upvote/downvote feature so you can help ‘train’ Obie. I’m also really excited about some of the newer NLP and the background AI that we are working on. I’d love to say more about that, but my business team won’t let me.
Later on, we’ll probably look into bringing Obie to other platforms: Facebook messenger, Skype, Telegram, Hipchat or Kik. The bot framework was designed to be fairly transferable to Facebook’s developer platform, but since companies are the primary buyer we’ll have to see if that makes sense.
