Messaging platforms are gaining more and more mindshare. Michael Wolf’s deck on the Future of Tech and Media shows that messaging will soon blow past social media as the dominant media activity and will add 1 billion new users by 2018. Given these numbers, the opportunities for messaging platforms are getting very interesting – particularly in terms of how to leverage messaging platforms to help people solve problems at work and in day to day life.

Improvements in Artificial Intelligence (AI) and Natural Language Programming (NLP) now make it possible to: 1) use conversational language as command line (the bot understands what you want) and 2) automate the execution of the command (the bot does what you asked for). Extending this to a messaging platform…it’s possible to text your request and have it fulfilled by an automated and scalable backend.

As the first stage, we’re already seeing messaging bots that act as gateways or aggregators to existing services. These are typically shortcuts to on-demand services, for example, to order food or an Uber. However, the potential for such aggregators is rather limited, since there’s not much value add and most on-demand services will likely have their own messaging bots in the near future. For example, 99designs has a Tasks Bot for Slack.

Growth areas for messaging bots

The real opportunity in messaging is to help people solve problems in areas where the underlying service doesn’t exist or the user doesn’t already know about the service (aka discovery).

The challenge here is you need a lot of data about customer preferences and behaviors. At the beginning, there’s a high level of human involvement needed to train the AI bot on what the user is really asking for and how to best solve their need.

BuzzFeed provides a look at the human training involved in developing M, the artificial intelligence that Facebook is building into its Messenger app. Currently, whenever someone sends M a message, the AI engine takes a first stab at the response, but this response is always reviewed, and then possibly edited, by a human trainer. All this data is fed back to M so that it will learn to get better and better. The idea is that in time, more and more tasks can be executed by AI and the service can scale.

Startups Operator and Magic are following a similar playbook. But it’s clear that any company needs a lot of funding to support all the human assistance while building the AI backend. On top of the funding challenge, there’s a real risk looming for startups who enter this space that Facebook or another platform will soon own the business.

Automating tasks in the enterprise

The other interesting opportunity for AI is to automate common chores for business users. We’ve already seen several virtual assistants go after the meeting scheduling market: e.g. x.ai and Clara Labs. The key here is removing any friction by being part of the regular workflow; for example, copy your virtual assistant on an email for an upcoming meeting and AI will take over from there.

It’s easier for startups to gain traction in the business space, since the tasks are relatively cut and dry. For example, a bot can quickly grasp how to schedule a meeting with predefined rules versus having to field consumer requests on anything from ordering an anniversary present to drawing a picture of the sunset. For this reason, you probably don’t have to years of stealth mode to build a solution catering to enterprise users.

In addition, the rise of Slack as a platform provides the opportunity to add further value: e.g. schedule internal meetings or collect regular updates from group members (see this good overview of the current Slack bots).

However, given the fact that external communication in the enterprise is still centered around email, it will be interesting to see how AI players will bridge the different communications platforms or if they will just focus solely on email, Slack or another platform. The other big question is if these apps will stay focused on narrow use cases or try to position themselves as a complete personal assistant.

The bottom line is we’ll be looking closely at the AI and messaging space in the coming year. The emergence of a new communication platform always sparks major innovation. Winning the consumer space will take heavy people-power and funding. There are easier opportunities to be had in the enterprise side, but you’ll need to figure out which use cases and which communication platform to pursue.