AWS embeds DuckDB into Aurora PostgreSQL and releases open-source AI decision model
Amazon Web Services Inc. has integrated the DuckDB analytical engine into Aurora PostgreSQL to enable faster queries of live and historical data without ETL pipelines. Concurrently, AWS released Strands Decider 2B, an open-source decision model optimized for rapid, low-latency AI workflows.
Editor, Lazyfounder

30 SEC SUMMARY
- AWS has embedded DuckDB into Aurora PostgreSQL, enabling direct querying of Apache Iceberg data lakes and Parquet files without ETL pipelines.
- The integration reduces network hops and speeds up analytical queries by processing scans within Aurora.
- AWS acquired DuckLabs B.V., the developer of DuckDB, in September 2026.
- AWS released Strands Decider 2B, an open-source decision model optimized for rapid, low-latency AI workflows.
- Strands Decider 2B is built on Qwen3.5-2B and achieves high performance in benchmarks like JevBench.
TABLE OF CONTENTS
KEY HIGHLIGHTS
- AWS embedded DuckDB into Aurora PostgreSQL to enable direct querying of Apache Iceberg data lakes and Parquet files.
- The feature eliminates the need for ETL pipelines, allowing seamless queries of operational and historical data.
- AWS acquired DuckLabs B.V., the developer of DuckDB, in September 2026.
- Aurora PostgreSQL now supports external Iceberg catalogs via AWS Glue federation.
- AWS released Strands Decider 2B, an open-source decision model for agentic workflows, available on Hugging Face and GitHub.
AWS embeds DuckDB into Aurora PostgreSQL for seamless data querying
Amazon Web Services Inc. has embedded the DuckDB analytical engine into its Aurora PostgreSQL database management system, according to SiliconANGLE. This integration allows customers to query Apache Iceberg data lakes and Parquet files directly from Aurora PostgreSQL without requiring data copying or ETL pipelines.
The feature enables seamless querying of both operational records stored in Aurora and historical data stored in Amazon S3. By processing analytical scans within Aurora, the integration reduces network hops and accelerates query performance. Customers can use existing PostgreSQL tools and endpoints to access combined datasets, simplifying application development and reducing infrastructure overhead.
AWS reportedly acquired DuckLabs B.V., the developer behind DuckDB, in September 2026. The capability supports external Iceberg catalogs through federation with the AWS Glue data catalog, and Aurora PostgreSQL versions 17 and 18 are compatible, starting with versions 17.11 and 18.6.
Aurora optimizes performance by filtering records during query execution and caching frequently accessed data. Developers can monitor metrics such as rows scanned, bytes read from S3, and cache hits. The feature is available in all commercial AWS regions at no additional charge, though customers pay for Aurora computing resources and S3 requests.
AWS launches open-source decision model for AI workflows
AWS has released Strands Decider 2B, an open-source decision model optimized for agentic workflows, as reported by SiliconANGLE and TechCrunch. The model is designed to make rapid, low-latency decisions by eliminating the need for text generation, which reduces latency in AI-driven processes.
Strands Decider 2B is built on the Qwen3.5-2B large language model but replaces its traditional LLM head with a 1-million-parameter 'pointer head' to deliver calibrated choices. The model achieves high performance in benchmarks like JevBench and can run locally with latency under 150ms, making it suitable for rapid experimentation and deployment.
The model was developed by AWS’s Strands Labs, an organization focused on AI agent tools. According to Marc Brooker, a Distinguished Engineer at Amazon Web Services, the project was inspired by TypeSafe’s Jev, a decision model released earlier this year. Brooker’s prototype briefly topped the JevBench ranking for models of its size.
Strands Decider 2B is available for download on Hugging Face, with its full codebase, training scripts, and examples accessible via GitHub. AWS encourages developers to use the model for tasks such as model routing, tool selection, context management, and guardrail enforcement in AI workflows.
Decision models gain traction for AI agent applications
Strands Decider 2B joins a growing category of decision models, also known as 'System 1 models,' which make rapid choices without generating text, code, or other outputs. Each decision is assigned a confidence score, allowing users to gauge accuracy. These models are distinguished by their speed and low latency, making them ideal for AI agent workflows where real-time responses are critical.
TypeSafe AI Inc. introduced Jev earlier this year, naming it after economist William Stanley Jevons to reflect its goal of leveraging the theory that falling costs of intelligence could drive increased demand. Dozens of similar models have since emerged, indicating broad interest in lightweight, high-speed decision-making tools for AI applications.
Despite the growing competition, TypeSafe’s CEO Diogo Almeida has stated that he does not yet see real competition emerging. Almeida has criticized the current wave of models as lacking true intelligence and dedication, suggesting that many are more focused on incremental improvements than breakthrough innovation.
What this means
Lazyfounder analysis — our interpretation, not reported fact.
AWS is doubling down on efficiency—both in database management and AI workflows. By embedding DuckDB into Aurora PostgreSQL, AWS eliminates a persistent pain point for operators: the need to duplicate data or maintain clunky ETL pipelines just to run analytical queries across live and historical datasets. This move isn’t just about speed; it’s about reducing complexity and cost, which aligns with the demands of startups and enterprises alike.
The release of Strands Decider 2B signals a broader industry shift toward specialized, lightweight AI tools. While LLMs dominate headlines, the reality is that many AI workflows don’t require the full power (or cost) of a frontier model. Decision models like Strands Decider 2B and TypeSafe’s Jev are filling a gap, offering fast, low-latency choices for agentic tasks. For founders building AI-driven products, this trend opens up new possibilities—especially for use cases where speed and cost efficiency matter more than generative output.
That said, the proliferation of these models raises questions about differentiation. If dozens of similar tools emerge, the challenge for startups will be identifying which ones genuinely improve performance without sacrificing intelligence. AWS’s open-source approach with Strands Decider 2B could accelerate adoption, but it also invites competition to iterate and innovate just as quickly.
Key takeaways
- Aurora PostgreSQL’s new DuckDB integration simplifies data querying across live and historical datasets.
- The integration reduces infrastructure costs by eliminating redundant data pipelines.
- Strands Decider 2B reflects AWS’s focus on lightweight, high-speed AI decision-making tools.
- Open-source decision models like Strands Decider 2B and Jev are gaining traction for AI agent applications.
- Developers can experiment with Strands Decider 2B locally, enabling rapid iteration for AI workflows.
Related on Lazyfounder
Sources
- SiliconANGLE · 2026-09-30
AWS embeds DuckDB in its PostgreSQL DBMS to speed queries of live and historical data - TechCrunch · 2026-10-01
Amazon releases its own Jev clone as decision models flood the web - SiliconANGLE · 2026-10-01
AWS debuts Strands Decider 2B, a first lightweight decision model for accelerate agentic workflows
This story is an original summary drafted with AI by Lazyfounder from the reporting listed above and checked by automated validation. Facts are attributed to their original publishers; sections marked as analysis are Lazyfounder's. Where a source is in another language, facts were machine-translated and quotations are reported, not reproduced. Read the original coverage via the links, and see our AI policy and corrections policy.
About the author
Editor, Lazyfounder
Tarun Mottlia edits LazyFounders, covering Indian startups, funding rounds, AI and product launches. Every story on the site is AI-assisted and checked against its cited sources before publication.
More stories by Tarun MottliaGet the LazyFounder Brief
Startup, funding and AI news in a five-minute read. Join the early-access list.


