Garman told WIRED ahead of the event that Amazon also plans to introduce a variety of tools to help customers explore generative AI models. He says generative AI models are often expensive, unreliable and unpredictable.
These include ways to enhance the functionality of smaller models with larger models, systems to manage hundreds of different AI agents, and tools to prove the chatbot’s output is correct. Amazon builds its own AI models to recommend products and do other tasks on its e-commerce platform, but it primarily helps other companies build their own AI programs. It acts as a platform.
Amazon doesn’t have a ChatGPT-type product to promote its AI capabilities, but its range of cloud services could give it an advantage in selling generative AI to other companies, said CEO and principal analyst at HyperFRAME Research. says Stephen Dickens of List. “The breadth of AWS is going to be interesting,” he says.
Amazon’s own line of chips will help make the AI software it sells more affordable. “Silicon is going to be a key part of hyperscalers’ strategies going forward,” Dickens said, referring to cloud providers that provide the hardware to build the most capable AI at scale. . He also noted that Amazon has been developing custom silicon for longer than its competitors.
Garman notes that more and more AWS customers are moving from demos to building commercially viable products and services that incorporate generative AI. “One of the things we’re really looking forward to is seeing customers move away from AI experiments and proofs of concept,” he told WIRED.
Garman says many customers are less interested in pioneering the frontiers of generative AI than in finding ways to make the technology cheaper and more reliable.
For example, a newly announced AWS service called Model Distillation can produce smaller models that have similar functionality to larger models, but run faster and at a lower cost. “Let’s say you’re an insurance company,” Garman says. “You can take a set of questions, feed them into a very sophisticated model, and use that to train a smaller model to become an expert on those things.”
Another new cloud tool announced today, Bedrock Agents, can be used to create and manage so-called AI agents that automate useful tasks such as customer support, order processing, and analytics. It includes a master agent that manages a team of AI subordinates, provides reports on how the team is performing, and orchestrates changes. “You can basically create an agent that claims to be the boss of all the other agents,” Garman says.
