Artificial Intelligence will bring new Revolution in Fashion Industry
CEO/Chief Data Scientist
Linear Squared is a revolutionary Sri Lankan company offering unique solutions to apparel industry by integrating Machine Learning and Artificial Intelligence. Its solution Capacity Squared is the world’s first intelligent capacity apparel planning tool. Powered by algorithms built on AI, it automates the planning function and thereby eliminates human error and biases inherent in the manual planning process.
WFB Bureau: When did your journey start? How did the idea of developing AI-driven capacity planning program come to your mind?
Sankha: I started my career as a merchandiser at MAS Holdings, Sri Lanka. However, I quickly shifted to analyzing data for a living. I was at Nielsen as a Manager of quantitative market research and then at Amba research (Now, Moody’s Analytics) as an AVP of quantitative financial research. Around this time, I realized that I needed to learn more if I wanted to do something innovative in the quantitative side.
I was at the University of Connecticut USA, from 2007 to 2013 where I earned my Ph.D. in statistics and thereafter worked for a large insurance company in Boston in advanced analytics. By this time, I was spurred by the entrepreneurial spirit that exists in the DNA of the USA, and wanted to come back to Sri Lanka and give it a try. Linear Squared was formerly incorporated in March 2015 with my good friend Rajith Munasinghe as the co-founder.
WFB Bureau: What was the motivation behind developing an AI solution?
Sankha:AI is widely associated with robotics and autonomous systems particularly in the developed world.
There are many applications of AI in a variety of these sectors, however apparel manufacturing happens to be one sector that is overlooked and underserved by the developers and innovators of the western world.
Further, being an industry primarily based in and around the South and South East Asian region it was difficult to expect this sector being served by the Silicon Valley.
WFB Bureau: Why did you choose apparel industry specifically for your solution, Capacity Squared?
Sankha:We practically stumbled upon this opportunity! With Sri Lanka being an apparel hub with many world leading apparel manufacturers, we had many use cases to study. We also had the talent. Therefore, we saw it as a natural fit and a good opportunity for value creation.
Sri Lanka is home to some of the world’s leading apparel manufacturers. Just by associating them, we came across this problem in their production process that we could solve.
Also, other countries in the South and South East Asian region though strong in apparel manufacturing didn’t possess the people knowledgeable and trained in AI technology to develop innovative solutions for the sector.
WFB Bureau: For the understanding of our readers, can you explain the meaning of AI in simple language?
Sankha: Generally speaking, any application that mimics cognitive ability can be termed broadly as AI. Technically, there is a class of algorithms which are called neural networks that try to mimic the human brain. Therefore, you’d find neural network algorithms at the core of any AI application.
WFB Bureau: What kind of algorithms have gone into the development of Capacity Squared?
Sankha:As mentioned, there is a trained neural network that tries to predict line efficiencies taking into account historical data on apparel styles and operators. There is also a cutting edge optimization algorithm handcrafted for the apparel sector which does the optimized style allocation.
WFB Bureau: Apparel shopfloor means lot of unannounced critical conditions such as delayed targets, worker absenteeism, etc. Is your solution able to consider such conditions?
Sankha:Yes. Capacity Squared does the optimization allocation within multiple parameters and constraints. When the parameters and constraints change it produces the optimization plan that adheres to these new requirements. The real advantage of Capacity Squared is that it can produce a new optimization schedule under 5 minutes where as a manual planner will have to spend hours to re-plan since a small change in requirements could have a cascading effect on the whole manufacturing process.
WFB Bureau: Is the solution cost effective? What kind of preparations/changes one need to have at their shopfloor?
Sankha:Preparations- Almost none. It is fully automated and syncs with order information through an ERP. Therefore, the backend is seamlessly integrated to the shopfloor. The initial factory configuration is supported by Linear Squared. So, there is minimum effort required from factory to implement. Plus, it’s user friendly interface makes planning as simple and easy as clicking a button. We also provide training. But it’s very easy to operate and requires very minimum ‘handholding’
WFB Bureau: Which companies have opted for the solution currently? What kind of results did it witness?
Sankha:We’re currently running trials with a leading apparel manufacturer in Sri Lanka and several more are in the pipeline to commence within the next 2 – 3 months.
So far, we have achieved savings of 5-10 line days per month which can translate into new capacity for more orders and thus more revenue.
WFB Bureau: What is the future roadmap of the company?
Sankha:Linear Squared operates in three sectors: Apparel, Telco and Retail. We are in the process of developing a portfolio of products within each of these sectors with the help of industry leaders locally. The short-term plan is to be present in the South East Asian and Gulf regions. We will also continuously expand our product portfolio and will also explore new sectors that we can add value to.