Saturday, July 11, 2026

AWS Database Migration Service now allows offline schema conversion for Microsoft SQL Server, removing the need for direct connectivity to source databases. Users extract metadata using standard commands locally and upload it to DMS for processing, achieving the same results as the connected approach. This capability bypasses security reviews, firewall changes, and VPN setups that typically delay migration projects.

• Eliminates need for direct DB connectivity, bypassing firewall and VPN setup delays.
• Ideal for strict security policies restricting external tool access to production SQL Server.
• Delivers identical conversion results to the online connected approach.
• Allows internal security teams to review human-readable metadata files before upload.

AWS has made change data capture for Amazon Aurora DSQL generally available, allowing real-time streaming of insert, update, and delete operations to Amazon Kinesis Data Streams. This managed feature captures database changes without requiring infrastructure management and is designed to have zero impact on database workload performance. Users can integrate these events into microservices, trigger Lambda functions, or deliver data to S3, Redshift, and OpenSearch via Firehose.

• Stream real-time DSQL changes to Kinesis for event-driven architectures without managing infrastructure.
• CDC captures inserts, updates, and deletes with zero impact on database performance.
• Integrate with Lambda, S3, Redshift, and OpenSearch via Amazon Data Firehose.
• Available in all regions supporting Aurora DSQL; no additional setup required.

A model identified as GPT-5.6 Sol Ultra has generated a PDF purportedly containing a proof for the Cycle Double Cover Conjecture. The document was shared via social media links and is available as a standalone PDF. The release has garnered significant attention on Hacker News, indicating high interest in this potential mathematical breakthrough.

• AI models are now attempting to solve open problems in pure mathematics.
• Verify the proof's validity independently; AI hallucinations in formal logic are a known risk.
• The Cycle Double Cover Conjecture remains a significant unsolved problem in graph theory.
• Community scrutiny on platforms like Hacker News is critical for validating AI-generated math.
• Treat AI-claimed proofs as hypotheses requiring rigorous peer review, not established facts.

AWS has introduced an Apache Spark troubleshooting agent for Amazon EMR on EKS, enabling data engineers to diagnose job failures using natural language queries. The agent automatically analyzes Spark History Server data, distributed executor logs, and cluster configurations to identify root causes like memory errors, data skew, and resource contention. It provides automated root cause analysis and PySpark code recommendations, eliminating the need to manually sift through logs. This feature extends the troubleshooting agent's coverage to include EMR on EC2 and EMR Serverless, accessible via the console or CLI.

• Diagnose EMR on EKS Spark failures via natural language instead of manual log inspection.
• Agent analyzes History Server data, executor logs, and cluster configs for root cause analysis.
• Receives automated code recommendations to fix issues like memory errors and data skew.
• Now available across all EMR deployment options: EC2, Serverless, and EKS.
• Access directly from the EMR on EKS console or via the AWS CLI.

Amazon SageMaker Feature Store now supports high-throughput batch ingestion via BatchWriteRecord, allowing multiple records across feature groups in a single request. Users can also discover stored records using ListRecords without pre-known identifiers. Additionally, the offline store now allows creating tables and databases with custom names for better cataloging.

• Use BatchWriteRecord for lower latency and fewer API calls during high-volume feature ingestion.
• ListRecords enables record discovery in feature groups without needing specific record IDs.
• Custom naming for offline store tables and databases improves data organization and cataloging.
• Batch writes target the online store to accelerate real-time feature availability for models.

Amazon SageMaker Unified Studio Workflows now includes 19 new operators for services like Amazon Bedrock, S3 Tables, S3 Vectors, and AWS Glue Data Catalog. These additions enable users to orchestrate tasks such as managing Bedrock guardrails and provisioning S3 resources directly within the visual workflow creator. By integrating these capabilities, the update reduces the need for custom integration code or switching between different AWS consoles. This expansion allows data workers to build more comprehensive workflows that include triggering MWAA Serverless runs and managing catalog metadata.

• No custom code needed for Bedrock guardrails, S3 Tables/Vectors, and Glue Catalog tasks.
• Visual workflow creator now supports 19 new operators across four major AWS services.
• Centralize orchestration by triggering MWAA Serverless runs directly from SageMaker.
• Reduced context switching between consoles for data engineering workflows.

AWS Database Migration Service Schema Conversion now supports AI agent automation through the AWS MCP Server, enabling integration with coding agents like Kiro, Claude Code, and Cursor. These agents can execute complete migration workflows using natural language, autonomously creating projects, browsing metadata, converting schemas, and generating assessment reports. A dedicated skill, dms-schema-conversion, provides agents with predefined procedures and operational rules to reduce trial-and-error loops during migration tasks.

• Integrate DMS Schema Conversion with IDE-based AI agents via the AWS MCP Server.
• Agents can autonomously run full migration workflows using natural language instructions.
• The dms-schema-conversion skill enforces best practices and reduces trial-and-error.
• Supports automated schema conversion, metadata browsing, and report generation.
• Extends automation to code objects like stored procedures with agent assistance.

This research paper introduces the Context Graph, a live relational data structure that models enterprise entities and their state transitions over time. It proposes a Delta Detection Engine to monitor changes and a Proactivity Scorer to rank insights by urgency and relevance. An LLM-powered Surfacing Layer then delivers these proactive recommendations to workers before they explicitly ask for them.

• Moves enterprise AI from reactive query-response to proactive insight delivery.
• Uses a Context Graph to track entity relationships and state changes over time.
• Delta Detection Engine continuously monitors for meaningful state shifts.
• Proactivity Scorer ranks insights by urgency, relevance, and user persona.
• Aims to boost productivity by surfacing actionable info before human requests.

Google has released a library of Agent Skills compatible with the Stitch MCP server, designed to integrate with major coding agents like Claude Code, Cursor, and Codex. These skills adhere to the Agent Skills open standard, enabling consistent interaction across different AI development environments. The release includes a quick start guide for installing the plugin suite globally or per repository via CLI.

• Enables cross-agent compatibility for Stitch design tasks using the open Agent Skills standard.
• Supports popular coding agents including Claude Code, Cursor, Gemini CLI, and Codex.
• Provides CLI-based installation methods for global or sparse repository plugin management.

This paper addresses the transition of enterprise LLM applications from prompt-driven prototypes to production-grade systems by introducing a harness-engineering approach. It shifts deterministic behavior into code, manifests, and validation artifacts, establishing a replaceable composition boundary while keeping source-backed claims as the authority for runtime answers. The authors validate this architecture using a public dataset of five Korean corporate groups to demonstrate improved traceability and auditability.

• Shifts deterministic logic from fragile prompts to structured code and schemas for reliability.
• Establishes a replaceable composition boundary to decouple orchestration from model logic.
• Ensures runtime answers remain grounded in source data rather than model hallucination.
• Validates the pattern on real-world corporate data to prove enterprise-grade auditability.

TencentCloud’s new open-source project implements a fully local, four-tier memory architecture for AI agents, eliminating external API dependencies. It combines symbolic short-term memory to condense tool logs with layered long-term memory that structures conversations into personas and scenes rather than flat vectors. Integrating this system with OpenClaw reportedly reduces token consumption by over 61% and boosts task pass rates significantly.

• Uses symbolic short-term memory to compress tool logs into Mermaid diagrams, reducing token overhead.
• Implements layered long-term memory to structure conversations into personas instead of flat vector piles.
• Fully local execution ensures zero external API dependencies for enhanced privacy and control.
• Integration with OpenClaw yields 61.38% token reduction and 51.52% relative pass rate improvement.
• PersonaMem accuracy increases from 48% to 76% through structured memory distillation.

CausalDS addresses the gap between symbolic causal reasoning benchmarks and standard data analysis evaluations by testing LLMs within realistic agentic workflows. The benchmark utilizes systematically generated synthetic causal structures rather than limited templatized variations, ensuring greater diversity in evaluation scenarios. This approach allows for a principled assessment of how agents handle data generation and causal inference simultaneously.

• Existing benchmarks lack realistic data analysis combined with principled causal structures
• CausalDS uses systematic generation of novel synthetic causal structures for diversity
• Evaluates LLMs as integrated agents performing both reasoning and tool use
• Fills the gap between symbolic reasoning and practical data science workflows

Amazon SageMaker HyperPod now allows AMI-based node lifecycle configuration for Slurm clusters using continuous provisioning. This update enables clusters to provision nodes with pre-configured software like Docker and Enroot without manual S3 script uploads. It simplifies the setup of production-ready AI/ML training environments by handling configuration automatically as capacity becomes available.

• Eliminates need to upload lifecycle scripts to S3 for continuous provisioning clusters
• Automates node setup with required tools like Docker, Enroot, and Pyxis
• Handles Slurm accounting, SSH keys, and log rotation via AMI-based config
• Streamlines scaling of Slurm clusters as capacity becomes available

Slack engineering has introduced an agentic testing framework that leverages AI agents to execute workflows based on user intent rather than rigid scripts. This approach allows tests to adapt dynamically to UI and system changes at runtime, addressing the fragility common in distributed systems. The strategy is designed to complement existing deterministic unit and integration testing layers rather than replace them.

• AI agents interpret intent to drive test execution, reducing reliance on brittle, hardcoded selectors.
• Tests adapt to runtime UI changes, improving resilience in distributed Slack environments.
• This approach supplements, rather than replaces, traditional deterministic unit and integration tests.
• Aims to lower maintenance overhead for end-to-end automation suites in complex systems.

The article argues that the best engineering tools disappear into the background, allowing developers to focus on logic rather than the mechanism. It contrasts this with tools that demand constant attention or introduce cognitive load through poor design. The core thesis is that tooling should facilitate flow state by removing unnecessary abstraction layers and decision points.

• Prioritize tooling that requires zero configuration for standard use cases.
• Reduce cognitive load by hiding complex features behind simple defaults.
• Measure tool success by how quickly engineers forget they are using it.
• Avoid tools that interrupt workflow with excessive alerts or verbose output.

Amazon DocumentDB now supports R8g.24xlarge and R8g.48xlarge instances powered by AWS Graviton4 processors and DDR5 memory. The 24xlarge offers 96 vCPUs with 768 GiB memory, while the 48xlarge provides 192 vCPUs and 1,536 GiB memory. These configurations are designed to handle high-concurrency transactional applications and memory-intensive operational workloads with higher throughput and larger in-memory working sets.

• Graviton4-based R8g instances enable higher throughput for DocumentDB workloads.
• R8g.48xlarge supports 192 vCPUs and 1.5 TiB RAM for massive in-memory datasets.
• Deploy via Console, CLI, or SDK by modifying existing clusters or creating new ones.
• Ideal for high-concurrency transactions and large-scale document processing.

AWS Config has expanded its library with 191 new managed rules, significantly broadening governance capabilities for AI and core infrastructure services. The update covers Amazon Bedrock, SageMaker, ECS, EKS, RDS, Redshift, S3, and CloudTrail, enabling automated checks for encryption, logging, and network security. These rules can be deployed individually or bundled into conformance packs to streamline compliance across supported AWS Regions.

• Covers critical AI services like Bedrock and SageMaker alongside core infra like EKS and RDS.
• Automates checks for encryption, public access, logging, and data protection best practices.
• Deploy via individual rules or conformance packs for scalable governance.
• Availability depends on the underlying AWS service being active in the specific Region.

AWS has extended availability of R8-series instances, powered by sixth-generation Intel Xeon Scalable processors and Nitro cards, to the Asia Pacific (Tokyo) and Europe (Frankfurt, Ireland) regions. These new instances offer up to 43% better compute performance per vCPU compared to the R6in and R6idn generations. The R8in and R8idn variants provide 600 Gbps of network bandwidth, marking the highest enhanced networking speed available among EC2 instances.

• R8-series instances are now live in Tokyo, Frankfurt, and Ireland regions.
• Expect up to 43% compute performance gain per vCPU over R6in/R6idn.
• R8in and R8idn deliver 600 Gbps network bandwidth for high-throughput workloads.
• Ideal for real-time big data, in-memory caches, and 5G Telco applications.

AWS Organizations now automatically attaches Service Control Policies (SCPs) that prevent member accounts from leaving or closing themselves when a new organization is created via the console. This default behavior eliminates the need for manual initial configuration, ensuring immediate protection against unintended account departures. The change targets enterprises migrating to AWS or establishing new multi-account structures, providing central security teams with enforced governance from day one.

• New orgs created via console get departure-blocking SCPs automatically.
• Prevents member accounts from self-terminating or leaving the org.
• Simplifies initial security posture for new multi-account environments.
• Central teams get immediate governance without manual SCP setup.

AWS has extended Amazon S3 Vectors support to its US GovCloud regions. This service provides purpose-built vector storage for AI agents, RAG, and semantic search at billion-vector scale. It leverages S3’s elasticity and durability while offering dedicated APIs for vector operations without infrastructure provisioning.

• S3 Vectors is now live in US GovCloud (US-East and US-West) regions.
• Designed for billion-vector scale AI, RAG, and semantic search workloads.
• Uses dedicated APIs to store, access, and query vectors without provisioning.
• Retains core S3 guarantees: elasticity, durability, and availability.

AWS has expanded Amazon MSK Replicator to handle data replication from external Apache Kafka clusters—including on-premises, self-managed, or other cloud environments—into MSK Standard brokers. This update complements existing support for MSK Express brokers, enabling seamless workload migrations and disaster recovery strategies. The service automates replication, removing the need for custom infrastructure or open-source tool configuration.

• Replicate from external Kafka (on-prem, self-managed, other clouds) to MSK Standard brokers.
• Simplifies migration of workloads to MSK Standard without custom replication code.
• Enables hybrid and multi-cloud data distribution using managed replication.
• Supports disaster recovery by using MSK clusters as failover or backup targets.
• Eliminates need to manage custom replication infrastructure or open-source tools.

AWS Security Hub now includes an active Network Scanning feature that probes resources directly from the internet to verify actual public reachability. Unlike previous configuration-based checks, this capability discovers live IP addresses, VMs, and load balancers across AWS and Azure environments. It identifies open ports and running services, generating specific findings for each reachable endpoint to confirm exposure.

• Active probing confirms real internet exposure, not just theoretical risk from misconfigured rules.
• Covers resources across both AWS and Azure environments in a unified view.
• Discovers specific reachable ports and identifies services running behind them.
• Generates Security Hub findings for each open port to aid in remediation prioritization.

Amazon RDS for Oracle has added support for Oracle Database 26ai, the latest Long Term Support Release. This update integrates Amazon Bedrock, enabling Select AI to generate SQL from natural language and offering access to models like Claude and Llama. It also introduces native AI Vector Search for storing embeddings and performing semantic queries directly within the database, eliminating the need for separate vector stores.

• Oracle 26ai LTS is now available on RDS for Oracle with Bedrock integration.
• Select AI allows natural language to SQL generation for developers and business users.
• Native AI Vector Search enables semantic and hybrid searches without external tools.
• RAG implementations can run directly in SQL, keeping data within the database.

Fleet impact: For Oracle ExaCC/RAC and RDS Oracle fleets, this release enables native vector search and RAG capabilities, reducing latency and data egress costs by eliminating the need for separate vector databases. Upgrade planning should prioritize testing Select AI prompts and validating vector index performance against existing workloads to leverage the new AI features effectively.

The pglayers project publishes minimal Docker images for 53 PostgreSQL extensions, built from scratch and containing only necessary binaries and scripts. Users can stack these layers onto the official postgres image using COPY --from directives, eliminating the need for compilation or apt-get in the final container. This approach results in smaller, immutable images where extensions are activated via standard CREATE EXTENSION commands.

• Eliminates build tools and apt-get from final images, reducing attack surface and size.
• Supports 53 extensions including pgvector, PostGIS, and pg_cron out of the box.
• Simplifies deployment by allowing stackable, pre-built extension layers via Docker COPY.
• No compilation required; extensions are ready to use after standard CREATE EXTENSION.

PDU is an open-source tool that reconstructs metadata and exports tables from offline PostgreSQL data files and WAL archives without modifying the original directory. It supports exporting entire databases, specific schemas, or individual tables as CSV for inspection or migration to another instance. The tool also facilitates targeted recovery of rows affected by accidental DELETE or UPDATE operations by parsing WAL logs.

• Recovers data from crashed PostgreSQL instances by reading data files directly.
• Exports offline database structures and content as CSV without server restart.
• Enables targeted row recovery from WAL archives for accidental DML errors.
• Operates without modifying the original data directory, preserving evidence.