Follow the User: What’s the Next App Platform?

Okay folks, its time for a little bit of Jon-stradamus prognostication!

As a nerd of a certain age, I’ve experienced quite a few tectonic shifts in the consumer operating system market. 

The OS wars have been raging for decades, and for many years, vendors competed by making big, sweeping changes on a regular basis. 

Do you remember the transition from Windows 3.1 to Windows 95? How about the release of Mac OS X, a ground-up replacement for the classic Mac OS, created after Apple famously acquired Steve Jobs’ scrappy startup, NeXT Computer? 

Operating systems provided a platform for applications, and users would pick their OS based on the app ecosystem.

A Walk Down the Web’s Memory Lane

In 1999, people started to talk about Web 2.0. Before long, web browsers had evolved into application platforms, thanks to innovation with HTML, CSS, and JavaScript. 

No longer did applications have to be built using OS-vendor platforms, and distribution became a snap. Rather than shipping CD-ROMs around the world, ISVs could host web apps in data centers, and customers could consume applications as a service. The SaaS model was a natural consequence of a disruptive technology. The huge growth in data center consumption by SaaS ISVs would eventually lead to Amazon’s creation of cloud computing.

These days, consumer operating systems are becoming increasingly irrelevant, as the shift to SaaS applications has sadly made native apps redundant. Users spend 90% of their day inside web browsers. Chromebooks are now the norm in schools, running an “OS” that is essentially just a web browser.

Disruptive technologies that change user behavior have a massive downstream impact, touching every aspect of computing, and creating entire industries, business models, and ways of using computers. We’ve seen it time and again, from Apple’s release of the iPhone, to the rise of modern web applications, and recently… something new.

I think you know where I’m headed. Say goodbye to Web 2.0, skip over the ludicrous Web 3 entirely, and take a look at what’s next - AI application platforms.

History, Repeating as Usual

A handful of years ago, most end-user computing happened in the browser or on mobile devices. 

Today, consumers are spending a dramatically increasing percentage of their time inside AI assistants like ChatGPT and Claude. As these assistants become agentic, they move from being read-only tools to a new paradigm for end-user computing. AI assistants are the next web browser—a platform upon which applications can be built, monetized, and delivered.

What does this mean, practically? 

I believe that, over time, as user attention continues to shift from browsers to AI assistants, applications will have to provide agents that extend their data and capabilities into AI assistants, allowing them to be interacted with using natural language. 

I do not believe that the endgame here is web applications that provide their own AI chatbots. Instead, I think that web applications will create standard agents that interface with consumers’ AI assistants of choice.

Now, why is this distinction important? 

Every app on the web is embedding AI chatbots into their UI, so isn’t that good enough? Plainly speaking: no. Why? Consider user behavior! 

Over time, users will gravitate toward their preferred AI assistant first, before ever consulting with a web app, search engine, or mobile app. This means that end-user applications will need to live within AI assistants, where they can be used in tandem with other application agents.

Consider a common use case: travel. 

Let’s say that I am planning a trip to visit my family in Atlanta. On the surface, it seems routine, but there are many factors to consider. Off the top of my head:

• Many different schedules, including my own work schedule, my immediate family’s schedule, and my extended family’s schedules.
• Flight selection, including personal airline preferences, frequent flier programs, seat availability, and price.
• Hotels, rental cars, and a slew of additional travel requirements.
• Payment methods, such as credit/debit cards and bank accounts.

This list is far from comprehensive, but it is highly illustrative. Currently, executing the trip would involve logging into a half a dozen websites, conversing with every traveler, and then determining the best course of action. 

In an AI application platform, agents from airlines, hotel chains, and calendar providers could converse with one another, coordinating amongst themselves to plan flights, accommodations, and schedules all from a single place - your AI assistant. Articulate your goals, and let the assistant formulate a menu of options before finally allowing it to handle the rest.