Singapore-based startup GreyOrange Pte. Ltd, backed by Tiger Global Management and Blume Ventures, has been trying to capture the automated parcel-management space, recently fuelled by the rise in global e-commerce. The company, which had raised funds in two rounds -- undisclosed sum in 2014 and $30 million in 2015 – has an Indian subsidiary based in Gurugram. In a conversation with TechCircle, the company’s country manager for India and Saarc (South Asian Association for Regional Cooperation), Vivekanand, who uses his first name only, said the startup was now focused on artificial intelligence robotics to redefine automation in supply chain. Edited excerpts:
Why did GreyOrange decide to work in the supply-chain segment?
GreyOrange has always been focused on creating products in the supply-chain automation space, especially those targeted at warehouses or fulfilment centres. We had assessed that the supply chain formed a big piece of any kind of business cycle as it involves a lot of processes. We saw a gap in that space that could be bridged by technology. The tech would simply help the company streamline its supply-chain processes by increasing efficiency, cutting down costs and reducing cycle time frames.
Having said that, you also have to understand that supply chain is deeply connected with consumer behaviour, which is changing today. People who are not on e-commerce are getting onto e-commerce platforms and people who are already on the platform, want shorter delivery time frames and better experiences. This phenomenon puts a lot of pressure on companies and their staff to deliver better experiences, which can be easily done with the help of robotics. Robots are also becoming cheaper by the day and, unlike human beings, don't need holidays, wages and other benefits.
What are the different kinds of products that GreyOrange has and are they all powered by artificial intelligence (AI)?
Our product range includes Butler, Profiler, Pick Put To Light (PPTL) and the Linear Sorter. The Linear Sorter is built on the backbone of tried-and-tested pneumatic arms technology with next-generation IoT (Internet of Things) sensors and advanced software, offering package-handling capability. It routes packages based on industry-defined sortation logic such as destination, volume, weight or any other custom logic based on business dynamics. The system is completely modular and scalable and comes in three standard sorting configurations of 1,500, 3,000 or 6,000 packets per hour.
The PPTL system is a technology that improves operators’ productivity and boosts accuracy. The directed picking, put-away and sorting processes help operators minimise errors, speed up the cycle and improve order-processing accuracy, while improving overall productivity. The modular architecture enables addition of more locations, SKU (stock keeping unit) types or number of units dynamically, as the business grows.
Coming to the Profiler, this product is a high-speed packet-profiling system designed to capture volumetric and dead weight, along with the dimensions of the packet in real time. The technology can be leveraged across logistics, e-commerce, retailers, consumer packaged goods and automotive industries to automate dimensioning and weighing process, while achieving throughput of up to 3,000 packets per hour.
But the most interesting one is the Butler, which is built on artificial intelligence and machine learning. While other products such as the Sorter also have some coded automation in them, the Butler can do a whole lot more.
Could you explain what Butler is and how it uses AI to increase automation in warehouses?
Butler is targeted at warehouse automation. Let me explain with the help of a contrasting example. In a typical e-commerce process, a consumer places the order, that order sparks a trigger for the system of the company, the warehouse sees the order coming in. Now typically what happens is the order becomes a part of the pick list depending on some criterion, and that pick list is given to staff. The staff members collect the items and bring it to a consolidation point. At that consolidation point, there is another round of sorting, after which the product is packed and shipped.
Now with Butler, we can provide a storage management as well as order-consolidation process. This means that, when the products are coming to the warehouses for storage, instead of the staff going to racks and storing the items, Butler brings the rack to the people. The racks are moved by Butler into positions that it sees most fit. Butler derives this intelligence from AI and machine learning. The system keeps on learning on its own about which goods are moving faster, depending on day, time, traffic, discounts and promotional offers. For example, an iPhone sale is being run by the company where it has offered free covers as well. Once such an order comes in, the system then reads it and once there is new inventory in the warehouse, it will start putting the phones and covers in the same racks. This is just the explanation for the in-bound process.
In the outbound process, let’s say the order comes in, the racks move to the person on their own. The person picks up the product and gives it for packaging to be shipped. This automates the entire process. Butler can also print the shipping tags that are to be pasted on the packets, making the process faster.
How does Butler work under the hood and how do you train its AI engine?
In order to train the AI, data are required from the client. Data points such as the number and kind of SKUs it will be handling, and their picture size. The system then starts reading into the daily activities, drawing out patterns, and makes intelligent decisions based on the learnings from the patterns. These decisions are like which rack is to be stored where, which product is to be placed on which rack and also which path to take to cover the shortest distance. There is also a designing aspect involved in the entire process. Butler comprises of many robots and software pieces called Grey Matter -- the brain of the platform. One such robot component is the mobile storage unit. These units have to be designed in line with the inventory of the warehouse, depending on the weight it has to handle.
In terms of all the robot components talking to each other and Grey Matter, we lay down a Wi-Fi network seconded by a redundant network running on Zigby protocol so that the entire system does not conk out if the first Wi-Fi layer is down for some reason. In terms of the robots understanding what they are doing, there are barcodes placed on the floor, racks and products. This helps the robot identify where it is going or storing the rack. Once the barcode is scanned, the location is stored in the thematic map built into Grey Matter for the warehouse. So once the order comes in, the robot remembers where it has put the product by matching the barcodes and goes and retrieves it for the picker.
Do GreyOrange products require infrastructure change in warehouses for deployment? What about protection for Butler from hackers?
Clients don't have to do huge infrastructure changes for deploying our products and this is why we are taking up a lot of brownfield (existing) projects, along with new projects. In terms of security, Butler can be put on an intra-network or private network. However, many clients would want to put it on a public network because they would want to access some information from remote locations. In such cases, it has to be protected by firewalls, which is the current practice across industry verticals.
Which product is bestseller?
This is a difficult question. In terms of newness or freshness, it will be the Butler and we have started offering a beta version of PickPal -- a product that will work with the Butler platform. Having said that, the sortation systems are the most popular ones, especially in countries like India. We have already installed 40 of them and more orders are coming in. So both products have different kinds of customers. Butler's reception has also been good because we have been receiving a lot of calls and queries as people are interested to know how the system works. There is already a site in India where Butler has been deployed. Japan-based Nitori, primarily engaged in the sale of furniture and interior goods, currently uses Butler. We have sold our products and solutions across a variety of regions such as Europe, Australia, Asia and West Asia (the Middle East).
What is the return of investment (ROI) on GreyOrange deployments?
Although there is no straight-forward thumb rule, but any customer who today invests in any tech, looks at ROIs. But there are a lot of variables around it. Whatever ROI discussions we have done till date, customers saw that by investing in the tech, they can break-even or get an ROI in a period of three to five years.