David Mack
Apr 20, 2016 · 4 min read

Silicon Valley is currently obsessed with chat interfaces. Magic lets you use any on-demand service by SMSing. Siri is on 200 million phones. Luka gives you restaurant reviews and more once you ask. Quartz tells you the news as a friendly conversation. Chat APIs and chat-bots as a service are popping up everywhere. Everyone’s making Slack integrations. Google and other companies’ AIs want to answer your email. Whatsapp sold for a $19 Bn. Snapchat’s worth $16 Bn. Facebook is trying to bring build a chat platform every service can run on top of. WeChat has already done it in China.

Chat interfaces promise to be extremely easy to use: we all know how to chat, therefore we all know how to use these apps straight-away. The minimalism of an empty white screen thrills designers. The novelty of the interface satisfies consumers’ demand for something fresh. Therefore chat interfaces are appearing in every area of tech.

If you’ve used any of these chat interfaces, you probably ran into a few problems. First of all, the chat-bot struggles to understand you. Hearing pained conversations with Siri is commonplace, after a few frustrated repetitions the struggling user gives up and opens the browser.

Secondly, it’s not clear what you can/cannot ask for. You have to blindly ask for things, see what comes back, and remember what seems to work. When using Siri it’s hard to know if she will directly answer your question, or direct you to a web search.

In UX terms, chat interfaces lack discoverability. Whilst chat-bots have mastered many skills, and are great at making cute comments when they struggle with requests, you’re still playing a blind game of darts with them.

Thirdly, chat is a bad way to display many types of information. Design is the art of conveying information efficiently. A well designed app communicates efficiently. When using Magic to order food or a car it’s hard to know what is going to happen because you lack the menus, summary tables and real-time updates an app would provide.

Finally, these interfaces are slow and awkward to use. When you open up Quartz’s news app, it’s awkward to tell it “I want to see that news article you shared with me yesterday” or to browse idly between different genres of news.

Ironically, chat interfaces bring us full-circle back to how companies operated before the technological revolution.

With so many problems and limited examples of success, it’s tempting to write-off chat interfaces as a passing fad. However, they offer a glimmer of the future.

The big benefit of chat interfaces is that their scope is unlimited. Traditional UX based products have tightly defined the scope of what they can do: it is exactly the set of screens the company designed. If you want to Uber a helicopter, you can’t: there’s no button for it.

Moving to chat interfaces removes UX design as a barrier to delivering services and experiences. If a company can deliver something, it’s ready to answer yes to requests for it.

In a chat future, companies can evolve their offering and grow faster than ever before.

There are a couple of key developments required to launch this chat revolution.

Firstly, machines need to become better at conversing. Currently it’s uncertain whether the machine will understand you. Machine conversation needs to become as good as asking a friend a question.

It’s very believable this will be solved soon. Deep learning and recurrent networks are progressing rapidly in their capabilities. We’ve just seen computers eclipse humans at Go and in various forms of medical diagnosis. Machines can caption images with surprising success. Within a decade machines may comprehend writing better than humans.

The second required development is to blend traditional UX with conversation. For example, your Amazon app will list all the packages on their way to you. When you need to know something not shown, e.g. whether the same item could be ordered again in a different colour, it’s as simple as typing that question beside the order. The app can help you formulate the question, suggesting the things it knows about and helping you refer to products in its catalogue.

If the conversational machine can reply with pictures, tables, videos, can add persistent items to your menus, then we’ve fully brought together the benefits of traditional UX and chat interfaces. The problems of discoverability and awkwardness are solved. The resulting interface would be faster and smoother than any current one.

These advances are all within our grasp, and unlock a thoroughly new paradigm of app and company. Apps can deliver both instant gratification and concierge luxury. Companies can pivot, grow, learn and test things even faster than before.

Looking further forward, mixing this with potential developments in general purpose AI (that is, algorithms that can learn any new skill just like humans can) paints a startling future: the scope of what a product can offer becomes fluid and ever growing, even unlimited. There is no longer a division between customer support and product features, instead it’s a continuum of service. A product that writes its own UX in response to customer needs is far-fetched yet tantalising.

Considering the potential of chat interfaces, it becomes clear why they are receiving heavy investment from large companies and prominent venture capitalists. Next time you use Siri, remember she’s just an infant.

David is CTO of SketchDeck, see more writing on their blog

SketchDeck developer blog

Delivering design at scale

Thanks to Charles Roesser

David Mack

Written by

@SketchDeck co-founder, https://octavian.ai researcher, I enjoy exploring and creating.

SketchDeck developer blog

Delivering design at scale

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade