Connect with us
Dr Swami Sivasubramanian, Amazon Web Services VP of data and AI, unveils new Amazon Bedrock functions and features. Photo: ARTHUR GOLDSTUCK

Artificial Intelligence

AWS re:Invent 2024: Amazon sets new AI agenda

The announcements in Las Vegas were so momentous, Amazon Web Services called on the CEO of its parent company to preside over the big unveils, writes ARTHUR GOLDSTUCK.

Amazon and its giant cloud computing subsidiary, Amazon Web Services (AWS), last week set a new agenda for AI in 2025. During the annual AWS re:Invent conference and expo in Las Vegas, the companies unveiled an extraordinary array of new AI products, features, and – most significantly – partnerships, which in combination position the company at the forefront of the AI revolution. 

So momentous were the announcements, AWS called on the CEO of its parent company to preside over the big unveils.

During an opening keynote on Tuesday, Amazon CEO Andy Jassy joined AWS CEO Matt Garman on stage to announce a new portfolio of Amazon AI models under the Nova brand. This set the scene for what is likely the most significant edition of the event since 2017, when AWS, under Jassy, first rolled out pioneering AI tools that set the AI agenda for the next five years.

At the time the Sunday Times had reported: “Many of the new product’s sound like science fiction, but within the next five years will be regarded as the new normal in artificial intelligence.” Five years later, ChatGPT was launched, brining AI into the mainstream. This week, AWS aimed at repeating history, unveiling a suite of generative AI tools under the Nova brand, along with dozens of additional AI tools and features.

Nova models are generative AI tools designed for text, image, and video generation, as well as multimodal capabilities that combine these inputs. Jassy emphasised the strategic significance of Nova, describing it as “state-of-the-art in performance, offering groundbreaking capabilities at 75% lower costs.” He highlighted the accessibility factor, stating that the models make advanced AI features attainable for businesses of all sizes. 

Matt Garman laid out AWS’s vision for democratising AI. He pointed to the Bedrock platform, which simplifies the integration of generative AI into business workflows. “Bedrock gives you all the best models, the right tools, and capabilities, and many of these features you simply can’t get anywhere else,” he said. Users will access Nova through Bedrock, which hosts Nova along with competing models like Meta’s Llama and Stability AI’s Stable Diffusion. 

Bedrock’s utility extends beyond generative AI. It includes features like Guardrails, a safety framework that allows developers to limit their AI applications to specific topics or tasks, ensuring responsible and reliable outputs. Garman gave a practical example: “For an insurance application, you might only want it to answer questions about policies, not give health advice or veer into unrelated areas.” 

In a keynote address on Wednesday, AWS VP of AI and data Swami Sivasubramanian announced major enhancements to the SageMaker platform addressed the complexity of training and deploying large language models. Over 140 new capabilities were unveiled, including HyperPod, a system designed to optimise the resource-intensive training of AI models by automating processes like capacity management and fault recovery.  This innovation is particularly relevant as models grow in complexity, often requiring billions of parameters to perform effectively. 

“Imagine needing to train a large language model for 30 days,” said Sivasubramanian. “HyperPod allows you to reserve capacity, recover from interruptions, and continue training without manual intervention.”

Beyond training, SageMaker now offers tools like HyperPod Task Governance, which dynamically allocates computing resources across multiple projects to maximise efficiency. Sivasubramanian described it as “a game changer,” enabling organisations to reduce costs by up to 40% while improving overall utilisation. 

While the announcements paint a compelling picture, their real-world applications were demonstrated through collaborations with leading global companies. Autodesk, a pioneer in design and manufacturing software, has used AWS’s AI tools to transform its operations. 

Autodesk chief technology officer Raji Arasu was invited onto the keynote stage to explain how AWS’s ecosystem had enabled the development of foundational models for complex 3D design tasks.

“We reduced foundational model development time by half and increased AI productivity by 30% while keeping operational costs steady,” she said. These advancements are helping Autodesk’s clients create sustainable infrastructure and innovate in fields as diverse as automotive design and ocean floor restoration. 

This focus on innovation was echoed by AWS enterprise strategist Phil Le-Brun, who told Business Times: “Data isn’t the new oil; it’s the new wine – it gets better with age, but only if you use it.” Le-Brun, who recently visited South Africa to see how local customers were using such services, gave Old Mutual and Standard Bank as major corporations that were leading the way in adoption of data-driven tools.

Not all companies were getting the message. Le-Brun highlighted the importance of breaking down silos within organisations to unlock the potential of data. Citing IBM research, he said that 79% of challenges in leveraging data stem from organisational and cultural barriers. 

Chris Erasmus, AWS South Africa country general manager, told Business Times that many local organisations needed to take “foundational steps”.

 “Many customers are chasing the big outcomes that AI can deliver but haven’t reflected on the basics data strategy, operating models, and security.”

Erasmus pointed to Capitec Bank as an example of a company using AI to enhance customer engagement and fraud prevention.

“One of the examples is where you can get reassurance through the app that who you’re speaking to is actually one of their members.”

Such innovations, he said, built trust and streamlined operations, demonstrating the practical value of AI in South Africa’s financial sector.  Retailers, too, were exploring the potential of AI to personalise customer experiences. Erasmus cited AWS’s Rufus, a new personalised shopping assistant announced on Tuesday, as a tool that could revolutionise the way consumers interact with retail platforms.

“Imagine being able to choose where to get your groceries delivered from, whether from a local franchise or a corporate store, based on real-time insights,” he said.

The country’s fintech and retail sectors have also emerged as leaders in leveraging AWS’s services. Companies like TymeBank, Pick n Pay, and Woolworths are increasingly adopting cloud and AI solutions to streamline operations and improve customer service. Erasmus said there were distinc t opportunities for leveraging AI in the region: “We have some unique challenges in our country, like fraud, but these tools allow us to address them in innovative ways, driving meaningful outcomes for businesses and consumers alike.” 

Skills development remains central to AWS’s mission in South Africa. The company has surpassed its global goal of training 29-million people globally by 2025, reaching 31-million this year, with significant efforts focused on addressing the skills gap in South Africa. Erasmus described these initiatives as transformative: “There’s such an opportunity around skills in South Africa. Our investments are aimed at closing the gap between the massive demand for digital expertise and the availability of those skills.” 

Sustainability also came to the fore as an ongoing focus. AWS has achieved a goal of running entirely on renewable energy well ahead of schedule and continues to innovate in reducing the environmental impact of its operations. “We committed to being net carbon zero 10 years ahead of the Paris Agreement, and we’re well ahead of schedule,” said Erasmus. 

These efforts are part of a broader vision that combines technological innovation with social and environmental responsibility. Jassy encapsulated this ethos during his keynote: “This isn’t just about technology – it’s about empowering businesses to build a better future.” 

Erasmus echoed this ethos: “The journey begins internally, by improving processes and efficiencies, but it ultimately culminates in delivering exceptional customer experiences. With the right support, the potential is immense.”

* Arthur Goldstuck is CEO of World Wide Worx and editor-in-chief of Gadget.co.za. Follow him on Bluesky on @art2gee.bsky.social.

Subscribe to our free newsletter
To Top