In today's flourishing e-commerce landscape, consumers are accustomed to comparing products and their prices across various shopping websites on their computers or mobile phones, which in turn drives rapid growth of the e-commerce industry. Commonly used search engines such as Google and Bing are not optimized for online merchandise search so their search results are often not up to user expectations. Users end up going to major shopping websites to search for products themselves, spending a lot of time and effort. BigGo, specializing in online merchandise search, offers a search engine purpose-built for online merchandise search. Featuring an enormous product dataset and real-time data refresh, it has become the go-to place where consumers check historical prices and compare product prices. Not only is BigGo a top-choice partner to most Taiwan-based e-commerce operators, but it has also successfully expanded into Japan and Southeast Asia.
According to BigGo founder and CEO Dewei Yen (Kevin), general-purpose search engines like Google and Bing have problems such as massive datasets, insufficient crawl depth, and inaccurate search results. For example, when doing a search by image using these search engines, users will mostly find photographs of internet celebrities in their outfits or social media posts with no information on the items that users are looking for. In contrast, BigGo's vertical search engine is specifically built for merchandise search. Its BIRSE image search narrows the search scope to shopping website URLs and merchandise prices and directly displays the item's historical prices to users. Moreover, with access to merchandise information regularly updated by e-commerce operators, BigGo ensures that the latest merchandise data is presented to users.
Actively developing two new services using generative AI
Founded in 2016, the BigGo price comparison website is now an all-in-one shopping assistant that consumers cannot do without. In addition to price comparison across e-commerce platforms, BigGo has recently added more services including BIRSE image search, social media comment check, price monitoring and product description generator. Like many search engines, BigGo also used natural language processing (NLP) to build its services helping users compare merchandise information as well as prices and check historical prices as well as cashback across leading e-commerce platforms.
Kevin noted that after years of development, NLP can be considered the predecessor of a large language model (LLM). Amid rapid AI advances, BigGo develops all its services based on AI. In particular, its BIRSE image search and product description generator are developed using generative AI. For example, built on top of BIRSE image search coupled with LLM, the product description generator creates text content for images that e-commerce operators provide. One of its most compelling features is that it analyzes the product in the image, suggests a suitable market price, and automatically generates production descriptions about the material, style, brand, specs, features, and use scenarios, helping e-commerce operators bring merchandise online with much less time and effort.
In consideration of system maintenance cost and service scope, BigGo chose to adopt cloud service when it first decided to incorporate generative AI. After comparing the technology and support service of various cloud service providers, BigGo selected Amazon Bedrock - a fully managed service from Amazon Web Services (AWS) that makes high-performing foundation models (FMs) from leading AI companies. With Amazon Bedrock, BigGo was able to bring its image search engine, product description generator, BigGo News, and other services online without a glitch.
BigGo News searches for news stories on the Internet and then selects some of them for the LLM to research using multiple search engines to capture more up-to-date and supplemental information (called AI Search). It makes AI write news articles like a journalist. Furthermore, AI Search also analyzes key opinion leaders' (KOL) unboxing reviews and social media comments. All these are aimed at delivering clear and easy-to-understand merchandise information to users, enabling a more efficient and pleasant shopping experience.
Enthusiastically participating in the Startup Terrace Kaohsiung AWS JIC program, BigGo secures more partnership opportunities
Being used more than 35 million times per month, BigGo has become a top partner to many e-commerce operators, including Shopee, Momo, Rakuten, ETMall, and Yahoo Shopping. Its latest product description generator piques e-commerce operators' interest. Some of them have begun to use the tool on a trial basis. Referred by AWS, BigGo has joined the Startup Terrace Kaohsiung AWS Joint Innovation Center (JIC) program, aiming to increase brand and service visibility while gaining access to additional resources and enhancing its technological capability.
According to Kevin, headquartered in Kaohsiung, BigGo does not have a lot of opportunities to network with other companies. Through BigGo's participation in the AWS JIC program, he hopes to connect with potential partners and increase partnership opportunities. More importantly, he hopes the endeavor will help boost the company's competitiveness and market influence. As a matter of fact, its participation in several match-making events has successfully bridged BigGo with e-commerce platforms including 91APP, Senao International, PChome, and myfone.
Results of BIRSE image search using a tassel bag image