Pomelo is revolutionizing the fashion industry with machine learning and AI

Machine finding out and AI have a multitude of programs in a variety of industries, but fashion could possibly seem like a curious section for their use.

E-Commerce is significant business enterprise in Southeast Asia (SEA). With motion restrictions coming and likely so frequently in the area, buyers have ever more moved in the direction of electronic signifies of retail, from quickly-moving shopper items (FMCG) to electronics, to style.

Even though segments like electronics are normally perceived to be well known, statistics clearly show that the trend section of e-commerce, surprisingly, will get the lion’s share of the multi-billion e-commerce pie. 

The SEA digital fashion business claimed a 22% GMV progress in 2020, with a projected market place quantity of US$ 18 billion for 2021. These amazing stats counsel a huge option in this e-retail sector for brick and mortar SMEs to digitalize their products and services to faucet on this increasing market place.

E-Commerce in Southeast Asia

SEA is a hotbed of on-line retail, with several e-commerce platforms serving distinctive marketplaces. Well known gamers involve Shopee, Lazada and Tokopedia.

All are or intending to provide the SEA market place at large, with Tokopedia not long ago merging with supply huge Gojek to form an e-commerce, shipping, and monetary products and services large.

In other places, smaller players, generally makes, have taken to marketing their solutions on-line thanks to simpler entry to clients, and cheaper operating costs.

Well-known Thai fashion manufacturer Pomelo entered the trend e-Commerce scene in 2013, offering a hybrid brick and mortar and on the internet marketplace vogue and lifestyle service to consumers in Thailand. They at some point expanded their electronic offerings to the rest of Southeast Asia. 

Leveraging the electrical power of significant knowledge

When it comes to merchandise that depend on unstable traits, these kinds of as fashion, desire forecasting is a important element to travel effectiveness in operations and administration. 

The manner field is infamous for struggling from demand uncertainty, particularly with a deficiency of historical info and ever-transforming developments. These usually final result in overstocked apparel, which suppliers may possibly be forced to market off at reduction, in get to make way for much more present-day developments on the cabinets.

Pomelo CEO David Jou, who co-launched Lazada Thailand, recognized the chance to leverage the large purchaser, product sales, and historic general performance advertising details gathered considering the fact that 2013.

Not too long ago, Pomelo released PRISM, “an stop to stop brand name alternatives platform that aims to present total methods for brands to scale their business”. The system presents a thorough suite of resources and services from merchandising, to analytics, to marketing and advertising, to even logistics to assistance smaller manner retailers accomplish improved economies of scale.

They also hire a group of details scientists who can enable brand names determine the ideal possible prospects by their a variety of technological options like big details.

Equipment understanding and AI drives impressive insights

In an unique job interview with CEO David Jou, Tech Wire Asia sought to better fully grasp the technologies powering a vital component of their newest platform — its desire forecasting provider.

In accordance to Jou, their demand forecasting service is based mostly on a proprietary device-studying (ML) engine that took 18 months to build. Equipment studying is a subset of synthetic intelligence (AI) that enables devices to process facts and execute automatic selections in a extra correct way. 

The motor plays seriously into supporting equally the buyer and supply chain sides. On the customer entrance, it delivers customized expert services primarily based on the user’s habits on their web site and mobile software, taking into account “hundreds of millions” of information details which are then fed back again into the motor. 

On the source chain aspect, the ML motor takes advantage of predictive analytics to examine details details from solutions and matches them to the movements of related products and solutions in other places in the planet in buy to precisely realize and forecast their reputation. 

Prism feeds its ML engine with details from the two their online e-commerce platform Pomelo, as perfectly as from their brick and mortar outlets around Thailand. This data will then offer need forecasting, which will tell inventory and merchandize preparing, as very well as products enhancement.

“We have finished a assortment of assessments, with pretty correct results north of 85%. We’re all about finding the finest high-quality details and figuring out which facts factors are the greatest driver of product sales.”, shared Jou.

Most of the system takes advantage of cloud computing from Amazon Net Services (AWS). Nevertheless, they have still to introduce edge computing into their electronic infrastructure.

On data privacy and cybersecurity

When queried about their cybersecurity and knowledge privacy procedures, Jou shared that they have a independent group that is effective on knowledge, with a target on guaranteeing the appropriate permissions are granted. 

He asserts that customer details is strictly acquired from their cell apps and on-line retail outlet, and does not track person exercise elsewhere, these as on social media or external internet websites.

Also, they anonymize buyer data prior to feeding them into their ML systems. Since Pomelo’s inception in 2013, they have but to deal with any data leaks or difficulties. When we asked who their present customers are, Pomelo declined to respond to. However, Jou did state that, down the road, some licensed circumstance research of clientele will be unveiled.

Equipment understanding and AI also powers the provide chain

On the offer chain end, Pomelo doesn’t own nor run factories, and as an alternative, buys from 3rd-celebration makers.

At this time, Pomelo does not prepare to integrate any online-of-factors (IoT) methods in their provide chain nor brick and mortar retailers.

“We benefit from a proprietary stack termed Henry to automate processes. It manages every thing from forecasting, prototyping, substance sourcing, and even automatic bidding with factory associates. We try out to digitize our operations as a great deal as we can, with our solution dependent on software”, he additional.

At time of creating, Jou shared that the organization intends to soon start throughout all markets in SEA, with a focus on onboarding trend and lifestyle brands, with a lengthy-phrase focus on promoting sustainably generated style.

He also indicated that there will be opportunities to bring on board make-up, cosmetics, and skincare manufacturers to the platform as well.