GenAI at the heart of AWS Summit

AWS has gone all in on GenAI – and so have its customers

AI took centre stage at the AWS Summit on Wednesday, with customers like Zilch and TUI setting out some early successes with AI innovation.

According to Tanuja Randery, Vice President & Managing Director EMEA who opened the keynote address at AWS Summit London, UK businesses have increased their adoption of GenAI by 31%. 64% attribute revenue growth to that decision, and 90% are showing efficiency and automation improvements.

After positioning GenAI as an inevitable evolutionary stage, Francesca Vasquez, Vice President of AWS Professional Services & Generative AI Innovation Center, urged the packed auditorium to:

"Stop thinking about this technology as something that can just generate poetry, and think about how it can bring deep reasoning into your applications. Imagine adding features that were unthinkable before due to the need for human like reasoning.

"This shift will fundamentally transform how we build products and businesses."

A case study in this transformation is Zilch – the world's first ad-subsidised payments network. Philip Belamant, Co-Founder and Chief Executive Officer, Zilch was on stage to announce a deepening of his company's collaboration with AWS on which his cloud native business was born.

"We use AI across fraud, credit, customer servicing and buyer intent prediction," Belamant said.

"Our team would initially take weeks to build a new underwriting model, regression test that model and deploy the model into line. Today with AWS SageMaker we can do it in days. And here's the really interesting part. We're currently ingesting more than 10 times the data for that model than we used to. Today, initial tests are showing us the predictability of this model is now almost twice as predictive as it used to be – and we're just getting started."

Belamant stated an ambition to "transform commerce as we know it."

At present the Zilch model depends on their 3.6 million customers telling the app where they want to shop.

"What if we knew where you were going before you told us?" asked Belamant. "And one day, what if we knew what would be valuable to you before even you knew? We're going to be combining GenAI and ML to surface that."

Francesca Vasquez just about managed to convey the vast scale of AWS's investment in GenAI.

"While many of you here today may not engage directly in building foundation models, our deep investments in this layer enhances the capabilities and efficiency of the services that we provide at higher levels," she said.

"Tools like SageMaker make it easy for our customers to build and train models with tools and workflow for the entire lifecycle. These workloads consume massive amounts of compute power, and to make GenAI use cases economical and feasible you need to run your training and inference on incredibly performant, cost-effective infrastructure that's purpose built for AI. All of these innovations come together to make AWS the best place to run GPUs in the cloud."

Vasquez went on to emphasise the AWS partnership with Nvidia and the company's own investment in custom silicon which should reduce the costs to customers of training and inference, in addition to significant efficiency savings, reducing the energy required to power these workloads.

In investment terms, Amazon has concentrated it's fire in the middle of the AI stack, although it does own a stake in Anthropic (as does Google.)

"Amazon Bedrock is the easiest way to build and scale generative AI applications with large language models and other foundation models," said Vasquez. "Customers in virtually every single industry are using Amazon Bedrock to reinvent their user experiences, products, processes, and bring AI into the heart of their business."

Showcasing Amazon Bedrock was Peter Jordan, Group CIO of TUI, who set out how AWS had been a pivotal part of TUIs mission to leverage cloud, data and now AI to redefine the personalised travel experience. The big challenge, Jordan explained, was content generation. At any point in time TUI has thousands of holidays for sale.

"But to create SEO, optimised content at scale that matches the tone of voice of the brand and is personalised is really a big challenge," he said.

"By utilising model stacking, we could solve a problem that we couldn't solve with a single model. We first used Llama 2 to create the content and pass it through Claude 2 to do alignment on tone of voice and fine tuning on languages.

"Amazon Bedrock gave us the enterprise ready, secure and compliant platform to scale generative AI to all parts of our organisation. Model choice is becoming really important, We can iterate over new models like Claude 3. We can build generative AI solutions really fast, whether that's inspirational search or cutting our call centre complaint time in half."

Jordan rejects the "start small" approach that many companies have taken to GenAI.

"Our vision for AI is not to start small, but to use AWS to really scale, accelerate, drive efficiency, and enrich the customers' experience. Personalised from inspiration to destination."