2016 was the year of the bots. But then the hype outpaced the progress of conversational bots in the market.

Despite the hiccups, conversational bots and messaging platforms still have the promise to be a major way users access digital services. But for these to reach their potential, we need to unlock the unique value in this new medium.

When Steve Jobs introduced the first iPhone, he demonstrated how you could now get the entire New York Times homepage on your phone. Given the limited mobile browsers of the day, it was incredible.

It wasn’t until later that we realized that a shrunk-down version of the Times wasn’t really what we wanted on our phones. We wanted more mobile-friendly UIs, sure. But what we really wanted were the Ubers and Instagrams of the world — services that were built specifically for mobile. Uber on desktop is fine, but Uber on mobile is transformative.

We are currently in the shrunk-down-NYT phase with bots. Most bots right now port functionality that exists in apps and websites over to this new medium — now you can hail an Uber or buy flowers via a chatbot. That makes sense; the top four messaging apps now have billions of users between them, even more than the top four social networks. Companies want to go where users are, and they don’t want to miss the boat on potentially huge trends.

But we should be spending the majority of our energy creating experiences that are uniquely suited to this new medium.

In exploring these new experiences, it’s important to distinguish between two different but related concepts: bots with conversational interfaces and products built on messaging platforms. Both are exploding in popularity. The former are agents that the user interacts with via natural language (typically via text or voice). The latter may make use of conversational interfaces, touch/tap user experiences, or a mix of both.

So what are conversational bots and messaging platforms uniquely good for? Here are five key scenarios that illustrate the different use cases.

Conversational bots

Some situations call for the simple, natural language interaction of a conversational bot.

1. Where there are many options, preferences, and settings

Visual UIs often struggle when there are many reasonable options. Think about the iPhone settings app, or Microsoft Office menus and toolbars. Imagine if you wanted to navigate to and select the right options for reserving a place for lunch near your work that’s affordable but upscale, and good for a group of eight. Asking for what you’re looking for in natural language would dramatically simplify these interactions. Emerging support and troubleshooting bots give promising examples, from bots that offer legal assistance to others that help schedule meetings.

2. Where being represented as an agent helps

Conversational interfaces are intriguing, in part because they mimic the way humans interact via natural language. As such, conversational bots can also be useful in situations where conjuring a human-like connection helps: when building trust, establishing rapport, or giving advice. You can imagine how a health coach like Vida benefits from being personable — you want to be able to respect and trust your coach, and even feel like you’re letting them down if you don’t follow through.

Messaging platforms

Tasks that require accessing multiple resources or that take input from multiple users can benefit by adding more ways to interact with the messaging platform.

3. Where a group is communicating

Messaging is inherently social. The dominant paradigm here is not iPhone, but WeChat — messaging makes multi-user the default. So use cases that involve sharing with friends, or collaborating with a group, stand out. Users can now chat and take action with less toggling between apps and less effort spent trying to describe current status to others. Group food ordering, as Apple demoed during their introduction of iMessage apps, is a great example.

4. Where the user doesn’t want to switch out of messaging

A lot of time is spent in messaging apps. Exiting the messaging experience to execute something you want to message about later anyway can get frustrating. In one example of a workaround, a lot of folks take a screenshot of a Yelp write-up so they can then add it as a photo in their messaging conversation about where to go for dinner. Messaging platforms are adding functionality like the ability to share a map of your location with others from within the app to eliminate the need for such workarounds. Expect much more such functionality, particularly for quick interactions that don’t require a complicated UX flow.

5. Where asynchronous responses are expected

Web pages and apps are typically designed to be fluid, synchronous experiences. You tap on what you want, and it appears as you wait. Search engine response times are measured in milliseconds. Messaging comes with different expectations, however. Sometimes you’ll be texting, and will wait patiently as those ellipses bounce for 10 seconds. Or you’ll ask a question and then switch out of messaging, hoping to get a response in the next few minutes. New products built on top of messaging platforms can take advantage of these differing expectations and offer asynchronous responses, whether that’s by creating and running a custom automated research job that takes a while to process or by inserting human actions and responses, as Magic does.

We needed to really internalize a mobile-first mindset to take mobile to the next level. To truly unlock the potential of conversational interfaces and messaging platforms, we need to similarly shift our thinking. We must consider which experiences are uniquely suited to being conversational, and adopt a messaging-first mentality.

For more reading on these topics, dive into some fantastic writing by Nir Eyal, Dan Grover, Ted Livingston, Ben Evans, Chris Messina, and Sarah Guo, among others.

Get more stories like this on TwitterFacebook