When it comes to predicting what is going to happen with bots in the enterprise, we need to look to the past because bots and AI are not new, they are just more readily available at scale and cost. Within the last year, almost every major brand has a bot. In fact, it’s hard to not find a brand that doesn’t have some form of experiment or live bot. The graphic below is just a snapshot of the companies who have some form of chatbot today. It feels a bit like the old app store days, when everyone needed to have app, now everyone needs a bot.

Most of these bots look the same. Very text message heavy, some with media, but most are what I call fit for purpose bots. They are narrowly defined, focused on marketing or outreach to extend the brand for interaction or to incent a purchase. I love what Casper has done as a great example of extending their brand with a bot. It is a fit for purpose bot, meaning it performs only a single function, in this case just something fun.

Most bots are deployed to the Facebook messenger channel. Messenger has over 300,000 bots with 8 billion messages as reported at the last Facebook developer conference. Far away they are the platform leader. And given the breadth of users, most companies decide that they will start with a messenger bot and just focus on a single channel. What’s interesting though is that within most of these fit for purpose bots, regardless of the original intent, many customers use them to ask for help. And that frustrates consumers when the bot can’t answer their question.

As users, we have been trained for the past 15 years with a search mindset. We see a nice big text box and are empowered to type whatever we want in there. And we do this because we know we’ll get a list of answers and can easily scroll and choose what will be the best one for us. We are in control and our expectations is that the search will get close to getting it right. For whatever psychological reason, our expectations are different for a chatbot - the answer’s got to be almost right or you’ll lose me.

It’s the same for internal company chatbots as well. We are seeing an explosion of simple bots at our clients for internal processes, but they are mostly siloed and don’t talk to each other. At Avanade, we’re beginning to take on the next evolution. What our IT team decided to do was build a single centralized bot. Instead of having an HR bot, a new joiner bot, a help desk bot, we have a single bot with many skills. So, think like the Amazon Echo skill store. Anyone in our company can build a bot, submit it to the skills store, and then as a user I can enable whatever skills are available. It’s my bot, performing lots of integrated functions enabling a more digital experience.

I can ask the bot for help and it'll show me several different skills that it can help me with. And instead of jumping applications, I can easily perform some specific actions like installing a printer. It knows that I'm in our Seattle office and identifies all the printers, listing the ones closest to me that I can click to install.

Over the next year, I challenge you to expand your definition of a bot. We delivered a bot experience for an insurance company in Europe for their customer service desk. This company had an existing customer service process for claims support. What we did was create a headless bot that would read the incoming emails, understand the intent, reply if it had an answer, or escalate it to a human. It’s headless because it’s a bot that performs with responses and actions that are machine to machine and do not involve human input.

Virtual agents aren’t easy. You need to make sure you have the right scope, access to the data via microservices, and are thinking through the experience and tie to your brand via the bot personality. The narrowly defined fit for purpose bots is easy. But to deliver fully integrated and seamless experiences, you’ve got to start breaking down the silos in your organization. We see most of our customers in two spots: did we pick the right place to experiment and are we getting the results? If you’re getting the results, it’s now time to figure out how to scale.

I’d be remiss if I didn’t talk about voice. Everyone keeps saying voice is coming. And yes, it is coming, but it’s not ready for a global full scale audience. We will see expansion with the personal digital assistants with new functionality, but if we go back to my previous analogy that as users, we’re coming from a search mindset, most of us are not thinking voice-first yet. Yes, the data shows high voice growth, but change is hard. The intermediary is the blend of mostly graphic and text, supported by voice. You can see that with Amazon’s Echo Show or what Google launched recently with their Smart Displays. You are going to see leading brands find ways to make images, plus video, plus data, plus algorithms and voice work for them.

My challenge to you is to think about the bots you are working on today and which ones you are trying to scale. Are you working in isolation? Are there ways to break down the silos and begin to provide a seamless experience for your customers and employees? And it’s upon us as business leaders to ensure we are doing right by the technology. If you are using artificial intelligence technology in your bot, do you have an ethics framework in place to ensure you are doing right by all users? I’m excited to see what happens over the next year. I’d love to hear your thoughts, please share your experiences and stories in the comments below.