Amazon Web Services (AWS) bets on cloud “building blocks” to empower the next wave of digital transformation, and generate artificial intelligence (AI) is positioned as an important foundation component as computing and storage.
In a keynote address at the recent AWS Summit in Singapore, Mylan Tom Sembkobek, vice president of technology at AWS, laid out a vision that enables accessible, scalable and secure cloud services to turn their imagination into reality.
“What if you could create something you imagine?” Bukobeck said. “On AWS, we provide you with building blocks to build something…and with them, we saw you build something amazing.”
An important part of Bukobek's address focused on a set of tools designed to make the generator AI more powerful and accessible. These include EC2 P6 instances with Nvidia's Blackwell chips and AWS' custom silicon, including EC2 P6 instances, including fourth-generation Graviton processors and Trainium 2 AI chips.
She says these advances are already “changering the economics of AI,” with partners like humanity using hardware significantly reducing training costs for Claude Foundation models.
At the heart of AWS' strategy is Amazon Bedrock, a platform designed to facilitate the development of generated AI applications. According to Bukovec, the platform not only offers AI model selection, but also provides the ability to customize models using search and augmented data (RAG) and the integration of automated inference into bedrock guardrails.
Customers tested AI building blocks
To highlight the real-world impact of this strategy, AWS highlighted several well-known customers on stage. Australia-based global design platform Canva serves more than 220 million people each month and aims to reach 1 billion users, and relies on AWS to promote global infrastructure and ambitious AI-driven capabilities.
“If the platform and tools are not trustworthy, we can't design the world,” Canva said in a video message. “With AWS, you can get all the right baseline features available when you need it, safe, reliable, and when you need it.”
Singapore-based gaming hardware giant Razer is using AWS for its AI-powered gaming services. Razer Chief Strategy Officer Li Meng Lee has introduced a new AI developer ecosystem and highlighted tools like the Razer Game Assistant.
Another tool, Quality Assurance (QA) companion, uses AI to detect bugs, performance issues and other QA tasks that can consume up to 30% of your game's development budget. “QA companions dramatically reduce time to market, improve game quality and reduce QA time by 50%,” Lee said, adding that the tool has been tested by major game studios and is available on the AWS Marketplace.
In a local example, Ong Chen Hui, assistant chief executive of InfoComm Media Development Authority in Singapore, detailed his collaboration with the Singapore Academy of Law to create the GPT-Legal. AI Tools is working on the monumental task of summarizing more than 15,000 court decisions to accelerate legal investigations. According to Ong, the model has been fine-tuned in Singapore's unique legal context and incorporates a set of safety features to combat AI hallucinations, including a 90% fact-check score requirement.
“Our system uses the emphasis on similarity to show exactly where each summary paragraph comes from, and also flags weakly demonstrated paragraphs and potentially hallucinating entities,” praised AWS infrastructure and generative AI expertise as a tool to enable the tool.
