By Bryan Kirschner, Vice President, Strategy at DataStax
Imagine getting a recommendation for the perfect “rainy Sunday playlist” midway through your third Zoom meeting on Monday.
Or a receiving text about a like-for-like substitute for a product that was out of stock at your preferred e-commerce site 10 minutes after you’d already paid a premium for it on another.
Or arriving late for lunch with a long-time friend and being notified that “to have arrived earlier, you should have avoided the freeway.”
We all expect apps to be both “smart” and “fast.” We can probably all call to mind some that do both so well that they delight us. We can also probably agree that failures like those above are a recipe for brand damage and customer frustration—if not white-hot rage.
We’re at a critical juncture for how every organization calibrates their definition of “fast” and “smart” when it comes to apps—which brings significant implications for their technology architecture.
It’s now critical to ensure that all of an enterprise’s real-time apps will be artificial-intelligence capable, while every AI app is capable of real-time learning.
“Fast enough” isn’t any more
First: Meeting customer expectations for what “fast enough” means has already become table stakes. By 2018, for example, the BBC knew that for every additional second a web page took to load, 10% of users would leave—and the media company was already building technical strategy and implementation accordingly. Today, Google considers load time such an important positive experience that it factors into rankings in search results—making “the speed you need” a moving target that’s as much up to competitors as not.
The bar will keep rising, and your organization needs to embrace that.
Dumb apps = broken apps
Second: AI has gotten real, and we’re in the thick of competition to deploy use cases that create leverage or drive growth. Today’s winning chatbots satisfy customers. Today’s winning recommendation systems deliver revenue uplift. The steady march toward every app doing some data-driven work on behalf of the customer in the very moment that it matters most—whether that’s a spot-on “next best action” recommendation or a delivery time guarantee—isn’t going to stop.
Your organization needs to embrace the idea that a “dumb app” is synonymous with a “broken app.”
We can already see this pattern emerging: In a 2022 survey of more than 500 US organizations, 96%of those who currently have AI or ML in wide deployment expect all or most of their applications to be real-time within three years.
Beyond the batch job
The third point is less obvious—but no less important. There’s a key difference between applications that serve “smarts” in real time and those capable of “getting smarter” in real time. The former rely on batch processing to train machine learning models and generate features (measurable properties of a phenomenon). These apps accept some temporal gap between what’s happening in the moment and the data driving an app’s AI.
If you’re predicting the future position of tectonic plates or glaciers, a gap of even a few months might not matter. But what if you are predicting “time to curb?”
Uber doesn’t rely solely on what old data predicts traffic “ought to be” when you order a ride: it processes real-time traffic data to deliver bang-on promises you can count on. Netflix uses session data to customize the artwork you see in real time.
When the bits and atoms that drive your business are moving quickly, going beyond the batch job to make applications smarter becomes critical. And this is why yesterday’s AI and ML architectures won’t be fit for purpose tomorrow: The inevitable trend is for more things to move more quickly.
Instacart offers an example: the scope and scale of e-commerce and the digital interconnectedness of supply chains are creating a world in which predictions about item availability based on historical data can be unreliable. Today, Instacart apps can get smarter about real-time availability using a unique data asset: the previous 15 minutes of shopper activity.
‘I just wish this AI was a little dumber,’ said no one
Your organization needs to embrace the opportunity to bring true real-time AI to real-time applications.
Amazon founder Jeff Bezos famously said, “I very frequently get the question: ‘What’s going to change in the next 10 years?’ … I almost never get the question: ‘What’s not going to change in the next 10 years?’ And I submit to you that that second question is actually the more important of the two—because you can build a business strategy around the things that are stable in time.”
This sounds like a simple principle, but many companies fail to execute on it.
He articulated a clear north star: “It’s impossible to imagine a future 10 years from now where a customer comes up and says, ‘Jeff, I love Amazon; I just wish the prices were a little higher.’ ‘I love Amazon; I just wish you’d deliver a little more slowly.’ Impossible.”
What we know today is that it’s impossible to imagine a future a decade from now where any customer says, “I just wish the app was a little slower,” “I just wish the AI was a little dumber,” or “I just wish its data was a little staler.”
The tools to build for that future are ready and waiting for those with the conviction to act on this.
Learn how DataStax enables real-time AI.
About Bryan Kirschner:
Bryan is Vice President, Strategy at DataStax. For more than 20 years he has helped large organizations build and execute strategy when they are seeking new ways forward and a future materially different from their past. He specializes in removing fear, uncertainty, and doubt from strategic decision-making through empirical data and market sensing.
