BigGo Speeds Up Generative AI Product Launch by 50% using Amazon Bedrock

Taiwan-based BigGo is a provider of search engine and price comparison technology. The company leads the local market with an eighty percent share and operates as a major player in Japan, as well as in Malaysia, the Philippines, Indonesia, Thailand, and Vietnam. In addition to its own search engine and price comparison website, BigGo’s technology supports numerous ecommerce businesses across the region.

As AI becomes integral to ecommerce, BigGo has begun to integrate it more deeply into its service offerings. The company’s vertical search engine specializes in merchandise searches and uses natural language processing to deliver precise and relevant results. Meanwhile, BigGo’s BIRSE (BigGo Image Recognition Search Engine) service for ecommerce websites uses AI-driven recognition technology to facilitate image searches. To save customers time and resources, BigGo developed two generative AI–based services. The BigGo Description Generator automatically creates product descriptions for items in an online store, while BigGo News provides stories and updates about products featured on customer websites.

When BigGo began developing its generative AI capabilities, it realized that daily token limits imposed by AI providers on API users restricted its ability to scale services. BigGo calculated that the token cap would reduce the number of news articles it could publish daily by half. Additionally, because some online stores have millions of products, multiple customers using the BigGo Description Generator at the same time could risk service disruptions. “We became a market leader because of our products’ performance, and we didn’t want to jeopardize service quality,” says Kevin Yen, co-founder and chief executive officer of BigGo.

As BigGo explored solutions, it identified Amazon Bedrock—a generative AI service on AWS—with fewer token limitations for the large language model it planned to use, Anthropic Claude 3.5 Sonnet. “Amazon Bedrock addressed the pressing need for AI infrastructure that could support our customers at any scale,” says Yen. Seeking to maintain its competitive edge by launching its services quickly, BigGo found that Amazon Bedrockseamlessly integrated with its existing infrastructure. “We could start experimenting immediately with Amazon Bedrock as our foundational AI technology,” Yen explains. “It completely eliminated the need for custom-built infrastructure."

The company launched the BigGo Description Generator and BigGo News Service within approximately three months of initiating development. Using the flexible computational power of AWS and unrestricted token usage, BigGo conducted large-scale experiments and advanced its generative AI applications without delays. “The simplicity of integrating Amazon Bedrock and its managed services helped us streamline development by about 50 percent, significantly accelerating our time to market,” comments Yen. By utilizing AWS services, BigGo also cut development costs for both generative AI applications by half. Moreover, the company avoided investing in and maintaining costly GPU-based hardware and minimized engineering expenses due to the accelerated development process.

Within months of launching the services in mid-2024, BigGo saw a 50 percent increase in user engagement on its website and developed new revenue streams through local ecommerce and retailer partnerships. The BigGo News application, which delivers breaking news on popular consumer goods, recorded a record 35 million monthly active users. Other companies using the service also reported a 30 percent increase in user growth. “With the flexibility of Amazon Bedrock with Anthropic Claude, we can produce twice as many news articles using generative AI, publishing 200 daily with support for multiple languages,” says Yen. “This capability is helping us expand into markets beyond Taiwan and Japan.”

The BigGo Description Generator, offered as part of the BIRSE service, reduces time spent on creating product listings by 80 percent, helping save costs and improve ecommerce efficiency. Companies upload product images to the web-based application, which swiftly generates descriptions that cover the brand, specifications, features, and use cases. Using other features of the BIRSE service, businesses can also access pricing guidance based on market analysis to optimize sales. “We’re a leader in our field, and adopting generative AI on Amazon Bedrock helps reinforce our position, future-proofing our offerings against the competition,” states Yen.

To further drive sales, BigGo has joined the AWS Joint Innovation Center program, which, among other benefits, connects the company with potential customers. “The program is helping drive growth by introducing us to a market segment we haven’t reached before—smaller ecommerce businesses with fewer than 5,000 products,” explains Yen. BigGo plans to expand its use of Amazon Bedrock with the Anthropic Model Context Protocol to develop AI agents for personalized shopping experiences. “Developing AI agents with Amazon Bedrock will transform our business from a price-comparison service to a comprehensive intelligent product assistant,” comments Yen. “This way, we’ll be able to engage customers earlier in their journeys and help them make better-informed decisions.”