The first official day of programming at Amazon Web Services’ annual tech conference, AWS re:Invent, has concluded. It delivered a steady stream of product news with an unsurprising theme: AI for the enterprise. This year, however, the focus is on upgrades that give customers greater control to customize AI agents. This includes one agent that AWS claims can learn from you and then work independently for days.
The conference, which runs through December 5, started with a keynote from AWS CEO Matt Garman. He emphasized that AI agents can unlock the true value of AI. Garman stated that AI assistants are giving way to AI agents that can perform tasks and automate on your behalf, and that this is where businesses are starting to see material returns from their AI investments.
While AI agent news promises to be a persistent presence throughout the event, there were other significant announcements. Here is a roundup of the key announcements from the first day.
AWS introduced a new version of its AI training chip called Trainium3, along with an AI system called UltraServer that runs it. This upgraded chip promises up to 4x performance gains for both AI training and inference while lowering energy use by 40%. AWS also provided a teaser, revealing that Trainium4 is already in development and will be able to work with Nvidia’s chips.
AWS announced new features in its AgentCore AI agent building platform. One notable feature is Policy in AgentCore, which gives developers the ability to more easily set boundaries for AI agents. Agents will now also be able to log and remember things about their users. Additionally, AWS will help its customers evaluate agents through 13 pre-built evaluation systems.
AWS announced three new AI agents called Frontier agents. One, named the Kiro autonomous agent, writes code and is designed to learn how a team likes to work so it can operate largely on its own for hours or days. Another agent handles security processes like code reviews, and the third manages DevOps tasks such as preventing incidents when pushing new code live. Preview versions of these agents are available now.
AWS is rolling out four new AI models within its Nova AI model family. Three of these models generate text, and one can create both text and images. The company also announced a new service called Nova Forge. This service allows AWS cloud customers to access pre-trained, mid-trained, or post-trained models that they can then further train on their own proprietary data, with AWS pitching flexibility and customization.
The ride-hailing company Lyft was among the AWS customers sharing success stories. Lyft is using Anthropic’s Claude model via Amazon Bedrock to create an AI agent that handles driver and rider questions and issues. The company reported that this AI agent has reduced average resolution time by 87% and has seen a 70% increase in driver usage of the AI agent this year.
Amazon announced AI Factories, a system that allows large corporations and governments to run AWS AI systems in their own data centers. Designed in partnership with Nvidia, the system includes technology from both companies. While users can stock it with Nvidia GPUs, they can also opt for Amazon’s newest homegrown AI chip, the Trainium3. This system addresses data sovereignty, meeting the need of governments and companies to control their data locally.