By Mukundha Madhavan, APAC Tech lead, Datastax
In the age of personalisation, we’ve collectively come to expect things to be tailored exactly the way we like them. From the perfect music playlist for every conceivable occasion, to substitute or complementary products for grocery runs; there’s no denying that apps being able to anticipate our needs is a major draw.
With it being more common for people to expect apps to be both “smart” and “fast”, we can all name at least one that is – and, similarly, the ones that aren’t. Often the latter sticks in our minds, with failures acting as a repellent – if not outright frustration.
The consequences of failing to meet these expectations can be a recipe for brand disaster. Organisations today are at a precipice for how they align their digital architecture in line with the twin goals of “fast” and “smart”. Here’s where organisations need to calibrate their definitions of both so that they can ensure their real-time apps are equipped with artificial intelligence (AI), while every AI app is capable of real-time learning:
Fast enough just isn’t good enough
Even five years ago, it was already evident that what app users thought was fast was outdoing what organisations were delivering.
Back then, the BBC found that for every additional second it took for a page to load, 10% of users would vote with their digits and leave.
Fast-forward to today, as Google considers load time when ranking search results. Add in the fact that speed is relative and constantly evolving, and it’s clear that organisations need to accept the fact that they can’t stop striving to be faster.
Real-time agility or bust
With AI becoming more accessible, the business landscape is looking for various means and ways to deploy use cases that drive growth. India is a unique example, in the sense that, as the NASSCOM finds, over two-thirds of locally based organisations’ AI investment constitutes less than 10% of overall tech investment.
This means those who act fast can be trailblazers and set themselves apart from the crowd. The potential this brings ranges from winning chatbots that satisfy customers to impactful recommendation systems that deliver revenue uplift. With apps expected to act based on insights at a moment’s notice, Indian organisations need to embrace real-time agility or customers will deem their apps broken and shun them.
Analytics-ready data in an instance
There’s a gulf that separates apps that are smart in real time and those that get smarter in real time. The difference is that the former relies on batch processing, which leaves some temporal gap between what occurs in the moment and the data driving an app’s AI. And this leaves machine learning (ML) models wanting, as historical data can be unreliable when the margins are finer. For example, a ride-sharing or navigation app would be widely panned if it relied solely on old data to predict traffic, instead of real-time data.
When the components driving the business move as quickly as they do, it becomes critical to do more than just the batch job when building apps. Therefore, the AI and ML architectures of yesterday just won’t cut it when customer expectations are evolving at light speed.
Doing that rests on being able to ingest millions of messages with ultra-low latency to process and stream high-volume, fast data flows in a distributed environment. This enables delayed batch jobs to be replaced by immediate updates that automatically stream data changes and provide an accurate picture that always reflects the current state of data. As a result, enterprises can simplify and modernise their data architecture, without the inefficiencies of batch processing.
Change is the only constant
As businesses look to the future, finding firm ground to build upon can feel disorienting. This is especially so in the face of the profound changes we’re facing today. India’s own ambitious USD1 trillion digital economy target by 2025 – driven in significant part by AI – highlights that continuous evolution is non-negotiable.
However, this does not preclude the possibility of developing effective business strategies to navigate the future; after all the tools to do so already exist. Enterprises can already do more than capture and warehouse data for future intelligence. They can position themselves to win by activating data in real-time to deliver in-the-moment experiences to customers, create instant intelligence, or fuel ML models. Those who with the conviction to act and harness these tools will come out trumps, those who sit still and wait risk becoming irrelevant.
