Official AWS What's New (AI-filtered) & AWS ML Blog RSS + GitHub + Hugging Face.
61 updates · page 1 of 5
Reflection AI introduced Beam, an open-weight model for agentic coding and long-horizon tasks such as maintaining code, using tools, interpreting results, and completing multi-step workflows.
Why it matters: A new model option to evaluate — compare quality, price, and context length against what you use today.
Web Search on Amazon Bedrock AgentCore is now in US East (Ohio), US West (Oregon), Asia Pacific (Sydney), and AWS GovCloud (US-West), giving agents current web grounding.
Why it matters: Agents in those Regions can ground research, summaries, and answers with managed web search.
Strands Agents SDK python v1.61.1 drops reasoning blocks a target model rejects when history crosses model families, and surfaces a subagent's stop reason.
Why it matters: Python Strands users can keep Bedrock history when switching model families without rejected reasoning blocks.
Amazon SageMaker Studio now lets new customers sign up with Google, GitHub, or Apple and land in a configured Studio environment without a traditional AWS account setup.
Why it matters: New AI and ML builders can open SageMaker Studio with an existing Google, GitHub, or Apple login.
Strands Agents SDK harness-python v0.3.1 includes the same Bedrock reasoning-block fix and surfaces a subagent's stop reason.
Why it matters: Harness users on v0.3.1 get the cross-family Bedrock history fix and a clearer subagent stop reason.
Meta's Muse Spark Open is now on Amazon SageMaker JumpStart, an open-weight coding model for agentic coding, multi-step reasoning, and tool calling on AWS.
Why it matters: AWS customers can run Muse Spark Open on SageMaker JumpStart for long-running coding agents.
Strands Agents SDK python v1.61.0 packages Strands skills for Claude, Codex, and Cursor, and adds a DecisionModel primitive and LLMDecisionModel.
Why it matters: Python Strands users can install skills for Claude, Codex, and Cursor and use the new DecisionModel types in v1.61.0.
Strands Agents SDK harness-cli v0.2.1 packages Strands skills for Claude, Codex, and Cursor, and adds a DecisionModel primitive and LLMDecisionModel.
Why it matters: CLI users on harness-cli v0.2.1 can install those skills and use the new DecisionModel types.
Strands Agents SDK harness-typescript v0.3.0 packages Strands skills for Claude, Codex, and Cursor, and adds a DecisionModel primitive and LLMDecisionModel.
Why it matters: TypeScript harness users can take the Claude, Codex, and Cursor skills plus DecisionModel in v0.3.0.
Amazon SageMaker Studio now lets new customers sign up with Google, GitHub, or Apple credentials and land in a fully configured Studio environment, with no traditional AWS account setup, credit card, or console navigation.
Why it matters: New AI and ML builders can open a configured SageMaker Studio from a social login without a credit card or console setup.
A guide shows how to build self-improving document processing with Amazon Bedrock AgentCore. It cites document-related costs of $19,732 per information worker a year, a 21.3 percent productivity cut, and staff spending up to 30 percent of their time on documents.
Why it matters: Teams receiving invoices from hundreds of suppliers can build AgentCore document processing instead of maintaining templates for every changing layout.
A walkthrough shows how to build an Amazon Bedrock AgentCore system that turns quarterly reports, compliance documents, and market analyses into client PowerPoint presentations.
Why it matters: Financial services teams can automate slide creation from those reports instead of reading hundreds of pages by hand.
Tamnoon built AI-powered cloud security remediation with Amazon Bedrock so findings from CNAPP tools, cloud providers, and posture scanners can move from detected to fixed.
Why it matters: Security teams can use the Bedrock approach to investigate context, find the owner, and remediate cloud findings that otherwise sit untouched.
A walkthrough shows agents paying for web resources that return HTTP 402 Payment Required, using Anchor Browser and AgentCore payments.
Why it matters: Developers can attach a payment path when an agent hits a dataset, API, or publisher that answers with HTTP 402.
Amazon Bedrock managed entitlements let a central account subscribe once to third-party model private pricing offers and distribute that access across an organization.
Why it matters: Teams with many AWS accounts can stop accepting each marketplace private pricing offer in every workload account.