Demystifying Machine Learning for Global Development

Dell, HP, IBM have all tried to transform themselves from being box sellers to solution providers. Then, in the world of Uber, many traditional products are fast mutating into a service. At Walmart, it is no longer about grocery shopping. Their pick and go service tries to understand more about your journey as a customer, and grocery shopping is just one piece of the puzzle.

There’s a certain common thread that run across all three examples. And it’s about how to break through the complexity of your end customer’s life. Statistics, machine learning, artificial intelligence can’t maketh the life of store managers at over 2000 Kroger stores across the country any simpler. It sounds way too complex.

Before I get to the main point, let me belabor a bit and humor you on other paradigms floating around. Meta software, Software as a Service, cloud computing, Service as a Software… Err! Did I just go to randomgenerator dot com and get those names out? I swear I did not.

The cliché in the recent past has been about how industries are racing to unlock the value of big data and create big insights. And with this herd mentality comes all the jargons in an effort to differentiate. Ultimately, it is about solving problems.

In the marketplace abstraction of problem solving, there’s a supply side and a demand side.

demand side | TO THE BRINK

The demand side is an overflowing pot of problems. Driven by accelerating change, problems evolve really fast and newer ones keep popping up. Across Fortune 500 firms, there are very busy individuals and teams running businesses the world over, grappling with these problems. Ranging from store managers in a retail store, to trade promotion manager in a CPG firm, a district sales manager in a pharma firm, a decision engineer in a CPG firm and so on. For these individuals, time is a very precious commodity. Analytics is valuable to them only when it is actionable.

On the supply side, there are complex math (read algorithms), advanced technology and smart people to interpret the complexities. And, for the geek in you, this is a candy store situation. But, how do we make these complex math – machine learning, AI and everything else – actionable?

To help teams/individuals embrace the complexity and thrive in it, nature has evolved the concept of solutions. Solutions aim to translate the supply side intelligence into simple visual concepts. This approach takes intelligence to the edge, thereby scaling decision making.

So, how do solutions differ from products, from meta-software, service as a software and the gibberish?

Meta-software | Service as a Software| | Mu Sigma

Fundamentally, a solution is meant to exist as a standalone atomic unit – with a singular purpose of making the lives of decision makers easy and simple. It is not created to scale creation of analytics.
For example a solution created to detect anomalies in pharmacy billing will be designed to do just that. The design of this solution will not be affected by the efficiency motivation to apply it to a fraud detection problem as well. Because the design of a solution is driven by the needs of the individual dealing with the problem, it should not be driven by the motivation to scale the creation of analytics. Rather, it should be driven by the motivation to scale the consumption of analytics; to push all the power of machine learning and AI to the edge.

In Anteelo you have a partner who can execute the entire analytical value chain and deliver a solution at the end. No more running to the IT department with a deck/SAS/R/Python code, asking them to create a technology solution. Read more about our offerings here.

error: Content is protected !!