I want to share my thoughts on bots this morning. As you’re reading this, please keep in mind that I’m not an expert on bots and we haven’t made any investments in the space. However, I’m still going to share my current thinking as I think that it may provide a good framework for how to think about the potential of bots in different areas.

The hype around bots has been ongoing for quite some time, and it hit a new peak with Facebook launching bots on Messenger last week. Bot enthusiasts claim that bots are going to be a new way for people to interface with online content. Rather than perform an action inside an app or a website, you’re going to do so by messaging a bot.

However, for this to actually take place, it needs to be easier to perform the action by messaging a bot than it is to do so yourself in an app or on a website. This is why it’s useful to evaluate the specific types of actions which bots could potentially help us perform to identify which, if any, of them meet this criteria.

Tom Tunguz from Redpoint Ventures wrote a great post about the different use cases for bots. Basically, the four use cases currently possible for bots are alerts, search/inputs, support, and bookings. Tom describes the use cases as follows:

“Alerts are notifications of news, events, account updates. Search means searching for movie showtimes or finding a customer record; input could be updating a CRM record. Support is talking to a vendor’s support team by chat. Bookings bots enable customers to buy products or services through chat.”

Let’s take each of these in turn.

I think that alerts are already very well-served by push (mobile and more recently web) notifications. I can’t imagine bots making these notifications any easier to consume.

The same is true for search and inputs. If I’m looking for something on the web or in an online database, doing so directly on that platform is pretty efficient. I have a tough time imagining how bots could improve it.

Bookings are a similar story. If I know what I want, it’s pretty straightforward to go to the website or app that delivers that product or service and buy it.

The only area where I believe that bots will improve the status quo is in support. Specifically, if I’m looking to perform a transaction but don’t know exactly what I want (pre-transaction support), or if I’ve already performed the transaction and am looking for post-transaction support, a bot could serve as a faster way to help me identify exactly what I’m looking for or get my questions answered about a specific transaction. It would achieve the former by quickly summarizing the alternatives available and grading each of them according to my stated preferences. This is something that the web and apps currently don’t do well. And it would achieve the latter by giving me faster answers to my questions than those that I could get from a customer service call center.

The reason why I see support as the most suitable use case for bots is because it requires two-way communication. All of the other use cases are examples of one-way communication, and the existing web and apps serve that need just fine. When there’s only one human involved in getting something done, it’s easy to take action and there are no communication problems. But when getting something done requires two-way communication, interacting with an intelligent bot could get that thing done faster and more accurately than if you were to interact with a human.

Other use cases where messaging a bot is more effective than directly performing the action yourself may emerge over time. But right now support looks like the best candidate.

Also published on Medium.