Amazon is hosting its annual user conference, re:Invent, this week in Las Vegas. The event brings together AWS practitioners, but also offers Amazon and its partners an opportunity to announce updates across the AWS ecosystem. Here are some of the highlights:
Custom transformations in AWS Transform
AWS Transform, a platform that can be used to modernize .NET applications, VMware systems, and mainframes, was updated with the ability to create custom transformations for organization-specific modernization tasks. Additionally, Amazon added pre-built transformations for a variety of common patterns, such as those found in Java, Node.js, and Python upgrades.
The company also added new capabilities for mainframe and VMware modernization, as well as a composability initiative for partners to integrate their own tools, agents, and knowledge bases to create custom workflows in AWS Transform for their customers.
Amazon Nova Forge
This new platform allows developers to build their own frontier models using Nova models. Users can combine their own datasets with Amazon Nova-curated training data, and then host their models on AWS.
“This data mixing approach significantly reduces catastrophic forgetting compared to training with raw data alone, helping preserve foundational skills—including core intelligence, general instruction following capabilities, and safety benefits—while incorporating your specialized knowledge,” AWS wrote in a blog post.
According to AWS, creating custom frontier models can be beneficial in a number of ways, such as building models that understand specialized processes or ones that understand brand voice and content standards.
Amazon Nova Act
Announced as a research preview earlier this year, Amazon Nova Act is now generally available. This new service helps developers build, deploy, and manage fleets of agents for UI workflows.
“Nova Act addresses the challenge of building reliable browser automation at enterprise scale. Powered by a custom Amazon Nova 2 Lite model, Nova Act excels at driving browsers, support calling APIs, and escalating to humans when needed. The service has core capabilities for web quality assurance (QA) testing, data entry, data extraction, and checkout flows,” AWS wrote in a blog post.
Nova 2 Lite and Sonic released
Nova 2 Lite is a fast and cost-effective reasoning model that supports extended thinking. Extended thinking allows the model to spend longer reasoning through a problem, but it is turned off by default for cost reasons. If enabled, it can be used with three different budget options to give control over the speed, intelligence, and cost tradeoffs.
Nova 2 Sonic is a speech-to-speech model for building voice interactivity. It provides several different expressive voice options, and also features improvements to speech understanding so that it can respond “not just to what the user said but how they said it.”
18 new fully managed open weight models added to Amazon Bedrock
The new models include ones from Google, Mistral, NVIDIA, OpenAI, Moonshot AI, MiniMax AI, and Qwen. These include the four newest models from Mistral, which are only available on Bedrock: Mistral Large 3, Ministral 3 3B, Ministral 3 8B, and Ministral 3 14B.
“With this launch, Amazon Bedrock now provides nearly 100 serverless models, offering a broad and deep range of models from leading AI companies, so customers can choose the precise capabilities that best serve their unique needs,” the company wrote in a blog post.
Amazon Bedrock AgentCore updates
New capabilities are being previewed that will help improve trust over AI agents deployed in production, including the ability to define clear boundaries for what actions an agent can take and built-in evaluations.
New capabilities for agents include long term memory that enables them to learn from past experiences, and voice functionality that allows users and agents to speak conversationally.
