Startdt (奇点云), a company focused on big data and AI, landed 24 million yuan (approx. US$ 3.8 million) in December 2017 in a Pre-A round led by Morningside Venture Capital, with contribution from Riverhill Fund, IDG and the individual investor LU Zhaoxi, former CEO of Alibaba. The financing will be used for team building and market expansion.
In February last year, the company raised a total of 16 million yuan (approx. US$ 2.5 million) from Riverhill Fund and LU Zhaoxi. It expects to soon launch a new funding round.
ZHANG Jinyin, founder and CEO of Startdt, joined Alibaba in 2004. He is the architect of Alibaba’s first data warehouse as well as the man behind Alibaba’s consumer information library and Alibaba Cloud’s big data and AI platform MaxCompute.
According to ZHANG, his company operates through four segments: data integration, business integration, unmanned retail and smart retail.
Because of ZHANG’s background in big data, Startdt started out focusing more on data integration services. A data integration platform, explained in simple terms, is a platform that integrates all of enterprises’ front-end data. And business integration is the linking of different businesses based on the integrated data. Take Alibaba for example. With databases for memberships, products, orders and transactions, the company is able to match a user account with different services.
“However, it turned out that it’s hard to offer a service like data integration at scale. The reasons are manifold. For one, the company must gain access to the varying systems of different enterprises and has to support private deployment,” noted ZHANG Jinyin. “Data and business integration are like two “clouds” which the company’s business was built on in the past. Now it’s complementing them with “terminals”, ZHANG said.
By “terminals” he means unmanned and smart retail outlets.
Unmanned retail refers primarily to unmanned stores and unmanned shelves, for which Startdt has rolled out two products respectively: “Startdt Magic Cube” and “Startdt Magic Box.”
The former is a mini-sized, cashier-free convenience store with an area of around 4m2, that is supposed to serve one customer at a time. Computer vision enables the store to let a customer in after scanning their face and is capable of detecting when goods are taken from and returned to the shelves. The goods are charged automatically upon closing of the door.
According to ZHANG, the operating cost of “Startdt Magic Box” is markedly lower than RFID-enabled stores, which require manually attaching RFID tags to merchandise. The downside is that the small space could limit the amount of merchandise displayed, but this can be addressed by assembling multiple cubes into one.
“Startdt Magic Box,” on the other hand, is an advanced version of vending machines. There is a vending machine for every 40 people in Japan. By this standard, China needs at least 2.6 million vending machines, but in reality, we have only 200,000, most of which are mechanical machines, ZHANG pointed out.
“‘Startdt Magic Box’, in contrast, is powered by computer vision technology. Customers scan their faces to open the machine, which they will be charged for the merchandise automatically upon closing,” ZHANG explained, who added that the combination of iPad, cameras and computer vision technology allows “Startdt Magic box” to operate at a cost 20% or even 10% the level for ordinary vending machines. Customers can purchase multiple items at a time and will not be charged if they’ve opened the door but haven’t taken anything.
ZHANG is confident that his company can produce the product at scale. “The relative difficulty of manufacturing ‘Startdt Magic Box’ compared to vending machines is like that of producing electric vehicles to conventional automobiles.”
In contrast to unmanned retail which involves no human staff, smart retail, in this case, encapsulates the congregated efforts to transform traditional brick-and-mortar stores with smart solutions. The transition to smart retail can currently be captured mostly in the retail sectors, including cigarette, alcohol, furniture and clothing. And the company also has its own answer to smart retail, the “Startdt Magic Mirror”.
The “Startdt Magic Mirror”, like “Startdt Magic Cube”, is enabled by computer vision technology. With this smart solution, shopping at a clothing store, for example, could take a whole new look. Other than leaving the customers themselves to sift through a daunting stack of clothes, the “Startdt Magic Mirror”, which has captured and stored essential information of the customer, such as face (labeled as the sole ID), gender, age, and the brands of the customer’s current wear beforehand through its camera, will automatically recommend to the customer some desirable clothes in tune with the customer’s current wear. The customers, by registering on “Startdt Magic Mirror”, will be rewarded with some coupons. Additionally, the company also brought the “pay with your face” solution to the retail stores’ traditional cashier systems.
To meet the varying requirements of different customers for the sales of hardware and software, Startdt introduced a very flexible sales approach: The customers can opt to buy the whole package, rent on a monthly basis or simply buy the hardware with the software being used on a SaaS basis. For facial recognition to work properly, high hashing power is essential. For this reason, the company opted for the server from Alibaba Cloud, which can also be customized based on an enterprise’s specific requirements.
At this stage, integration services and retail terminals each contribute 50% to Startdt’s revenue, said ZHANG. The company, in the future, will tilt more towards unmanned retail and smart retail. In addition to the retail enterprises that seek innovation with smart solutions, the construction of characteristic towns driven by the government is also a revenue source for the company, which currently accounts for 30% of the company’s total revenue.
The company is said to have since its inception in December 2016 inked deals with over 100 customers, which span the retail sectors, including mother & baby, clothing, cigarette and alcohol, furniture, finance, convenience stores and supermarkets, and commercial real estate.
Startdt currently has no more than 150 staff, among which 70% are engineers. The members in its startup team are all top talents who had previously worked with the giants, including Alibaba, Ant Financial, IBM, Baidu and Huawei.
LIU Ying, co-founder and COO of Startdt, had previously served as the spokesperson for the innovative services of Alibaba Cloud’s big data business and worked in IBM as the assistant for its global vice president and the general manager overseeing the sales and operation in China. SHI Jian, another co-founder and CTO of Startdt, is the former founder of the data team of Alibaba Cloud and the former supervisor of the Business Intelligence Department of Alibaba Cloud.