We at SocialToaster recently unveiled it’s newest tech software creation: the Artificial Intelligence-driven marketing content distribution bot ContentToaster.
With ContentToaster currently in it’s beta testing phase, we wanted to get a deeper look at the technology behind our latest game-changing advocacy marketing feature. So we cracked our GCal and put some time on Chief Technology Officer Ted Bever’s schedule (which was surprisingly easy). Ted leads the product team at SocialToaster, and is responsible for product roadmaps and the strategic technology vision and planning. He also gets in the weeds and work on the SocialToaster platform as well.
ST: How did you get started in programing?
TB: I guess my first programing experience came in middle school. I don’t know if, did you ever have a graphing calculator? So, I had the TI82, which supported a version of basic programming language. And I just remember thinking how cool it was that, you know, I could make a game or you know, solve a problem using what I was telling the computer, the calculator in this case to, to do.
I started college in the year 2000, at the University of Rochester in New York so, you know, worldwide web usage, it was really big. It was actually the start of the.com bust. So even though all that was going on, I was really comfortable working with computers, I was interested in what else I could do with them. And even though all that was going on with the markets and all, there definitely seemed like there was value in being able to learn how to code and knowing the technology there. So you saw it, you saw the advantages to it, and you took hold of the future just kind of knowing that that would be the direction that the will is going to take anyway.
I went up to New York for college and came back to Baltimore afterwards. After I came back, I actually interviewed with Brian (Razzaque, SocialToaster inventor and CEO), to be a developer at his previous company, the one prior to social SocialToaster, which was called VMT. It was a web services and website development shop. So I worked with him for a number of years before the idea of SocialToaster came about.
ST: Is what you’re doing now very different from what you were doing then?
TB: Yeah, they’re very different. Back then I was just a developer, so I was involved in writing code and completing client specifications and things like that. But now I’m involved in research and development for the things that we want to complete for the year and meet our business objectives with those. I’m involved in much more strategic things now.
ST: What were the early days of SocialToaster like as you transitioned from one company to the other and as your role changed to lead tech developer?
TB: It was interesting, and back with VMT there wasn’t like an agency-type situation where we were necessarily doing all sorts of social media strategy for our clients then. But I think that’s where Brian first saw that these businesses, these organizations, had these social followings and they didn’t know exactly what to do with them. So that was kind of the genesis of the idea for SocialToaster. So it was basically a side project for us at that time. It was interesting because we could only spend a limited amount of time on it until there was this opportunity to get outside funding. Once we could do that, we had the opportunity to move full time to SocialToaster.
ST: Now you’re working on a new project called ContentToaster. Can you explain what that is in layman’s terms?
TB: (laughs) Yeah, so I’ll try to give you the explanation that I gave to my parents. ContentToaster is an automated process for matching content to the users who are most likely to share it. That’s what it is. How we do that is we generate predictions around the content that enters into our system and based on what we’ve seen with other content that’s come through, try to see what the relationship is between that.
ST: And ContentToaster uses Artificial Intelligence?
TB: Yes. It uses AI tools to generate those predictions.
ST: How did this idea to use AI to match content to users come about?
TB: I can’t claim to have to come up with this idea. This is Brian’s idea; I think he identified an opportunity in what SocialToaster had prior to ContentToaster. We had targeting tools built in that were based off of information that was collected at signup or things that you might ask during sign up, like their content interests, topics they were interested in, basic demographic information, and you can use those things in the SocialToaster platform to target groups of users, but that doesn’t necessarily allow you to know if they’re likely to share the thing that you’re giving them. And there’s also the work that’s required to generate that piece of content within SocialToaster as well. But that sort of targeting took multiple steps. It could be a fair amount of work for someone to do and to understand and to get results. So I think Brian saw this idea of having a more automated process of pulling in content, matching it to users, and sending it out without having any interaction from the client at all. So that was the original idea, and our goal is to get that in place.
ST: As you’ve gone along those paths, to try and make this vision become a reality in the system, what have you learned?
TB: I guess I would say I’ve learned that automation is hard. That a lot of times replacing that set of eyeballs is not easy, and this kind of goes towards some feedback we’ve had from some of our clients who are using an early version of ContentToaster. This particular post, it didn’t look quite right and it would’ve been nice if someone had seen that before it went out. But that that’s just part of the learning process of, “Okay, we’ve identified this as a problem, now how do we adjust our algorithm to prevent it?”
ST: As you’ve been working on this, and especially now that you have seen some feedback, how do you hope that ContentToaster achieves or changes how people interact with the SocialToaster program?
TB: I guess on the fan side, I would hope that our users find that they’re getting content that’s relevant to their interests and keeps them engaged and participating in the program. As for the clients, it kind of goes back to making things easy for them and we want to make sure that they did get visible results and ultimately that they’re getting added value.
ST: You said that ContentToaster is still in early access. Do you have a date for when the full program would go live?
TB: We’re still in the process of collecting feedback and adjusting the algorithm as we see fit, so there’s not currently a timeline for when we’re gonna make it available to all clients; however any client has the ability to request to have ContentToaster turned on by contacting their SocialToaster account manager.
ST: What type of clients have been requesting early access to ContentToaster?
TB: At the end of the day, there are clients who aren’t a good fit for this level of automation. Some organizations may want a level of control that they wouldn’t really have with ContentToaster. So we see that this will be a benefit primarily to those organizations where they don’t have the time or the resources to put towards generating content specifically for SocialToaster and just want to be able to connect their existing social media networks and blog and have ContentToaster send that content out on an automated basis for them.
ST: What do you think the future is for ContentToaster? Will it just just stay at this iteration, or does it evolve into something else?
TB: I would say our current goals are to use our clients to gather feedback, to gather data and to use that to shape the algorithm itself so that we can make sure that we’re delivering optimized results for our clients. As I mentioned, we’re still looking at other ideas to enhance it. Like right now it’s sent out on a fixed schedule. We may want to consider having varied set times based on when users are more likely to share. We’re also right now collecting feedback more or less on a manual basis. I think we may want to incorporate feedback that doesn’t involve us having to review the form data submission, so something that allows you to spit feedback automatically back into the system. And we can use that to shape the results of that the system is spitting out.
Editor’s Note – This interview was edited for brevity and ease of reading; no material information has been edited.