Shareable cards

Thursday, July 09, 2026 — top 10 database + AI stories, one card each. Save the image, copy its caption, post.

Phone: tap Share image, pick Save Image in the sheet (LinkedIn's app won't take images from Safari directly), then tap the button again — it opens LinkedIn: start a post, add the photo on top, paste the caption. Desktop: Open → save the image, attach it in the composer, paste the caption.

Amazon Aurora DSQL CDC now generally available for real-time streaming

Open image

Amazon Aurora DSQL CDC now generally available for real-time streaming

▸ Stream DSQL changes to Kinesis for event-driven microservices without managing infrastructure.
▸ Trigger Lambda functions or load data into Redshift/S3/OpenSearch via Data Firehose.

Summary + source: https://offnetchat.com/news/2026-07-09/s/48d56a6a79b4ec51.html

#Databases #TechNews

PostgreSQL: Achieving Partition Pruning on Non-Partition Keys

Open image

PostgreSQL: Achieving Partition Pruning on Non-Partition Keys

▸ Partition pruning is not strictly limited to the partition key column.
▸ Data patterns and outlier handling can enable pruning on other fields.

Summary + source: https://offnetchat.com/news/2026-07-09/s/7746a1067927bf94.html

#Databases #TechNews

AlloyDB GA: Proxy Models Replace External LLM Calls with Local Inference

Open image

AlloyDB GA: Proxy Models Replace External LLM Calls with Local Inference

▸ Eliminates network latency and API costs by running inference locally inside the database engine.
▸ Achieves reported 2,400x throughput gains via smart batching and local processing.

Summary + source: https://offnetchat.com/news/2026-07-09/s/f14364ea5975dca1.html

#Databases #TechNews

Hugging Face integrates native vLLM backend for Transformers

Open image

Hugging Face integrates native vLLM backend for Transformers

▸ Eliminates need for separate vLLM server processes by embedding backend directly in Transformers.
▸ Leverages vLLM's PagedAttention for improved memory efficiency and throughput.

Summary + source: https://offnetchat.com/news/2026-07-09/s/f2a3364a10eb9acc.html

#LLM #TechNews

PostgreSQL enable_partitionwise_join: Decompose Big Joins

Open image

PostgreSQL enable_partitionwise_join: Decompose Big Joins

▸ Enable partitionwise_join to let the planner decompose joins into per-partition pairs.
▸ Both tables must be partitioned on the join key for this optimization to apply.

Summary + source: https://offnetchat.com/news/2026-07-09/s/6a6a21bc9717c3e2.html

#Databases #TechNews

Research: Orchestration Design, Not Model Size, Drives Enterprise Agentic AI Token Costs

Open image

Research: Orchestration Design, Not Model Size, Drives Enterprise Agentic AI Token Costs

▸ Token efficiency depends more on orchestration logic than the underlying foundation model chosen.
▸ Stop scaling token limits blindly; optimize context assembly and tool delegation first.

Summary + source: https://offnetchat.com/news/2026-07-09/s/1cb54419ad00aa6b.html

#AgenticAI #TechNews

Agentic Data Environments: Active Substrates for Safe AI Automation

Open image

Agentic Data Environments: Active Substrates for Safe AI Automation

▸ Agents require execution substrates beyond traditional databases to manage files, APIs, and state.
▸ New frameworks aim to amplify agent capabilities while strictly bounding failure costs.

Summary + source: https://offnetchat.com/news/2026-07-09/s/8f381243b8bd98f3.html

#AgenticAI #TechNews

Multi-Teacher On-Policy Distillation Causes Invisible Behavior Shifts

Open image

Multi-Teacher On-Policy Distillation Causes Invisible Behavior Shifts

▸ Multi-teacher distillation improves recall but risks over-calling tools
▸ Aggregate loss metrics fail to detect invisible behavior shifts

Summary + source: https://offnetchat.com/news/2026-07-09/s/55e2c6150ed97f84.html

#AgenticAI #TechNews

DeLS-Spec: Decoupled Long-Short Contexts for Parallel Speculative Drafting

Open image

DeLS-Spec: Decoupled Long-Short Contexts for Parallel Speculative Drafting

▸ Avoids costly full retraining of draft models by keeping DFlash fixed as the long-context expert.
▸ Introduces a lightweight local head to enforce intra-block causal conditioning in block-parallel drafting.

Summary + source: https://offnetchat.com/news/2026-07-09/s/85c1ea732297a7af.html

#LLM #TechNews

PALS: Layerwise Sparsity via Activation Percentiles Beats Uniform Pruning on LLaMA-2

Open image

PALS: Layerwise Sparsity via Activation Percentiles Beats Uniform Pruning on LLaMA-2

▸ PALS uses 99th percentile activation magnitudes to set per-layer sparsity, improving LLaMA-2-7B perplexity by ~0.96 over uniform Wanda.
▸ Gains are architecture-dependent: LLaMA-3-8B sees marginal improvement, while Mistral-7B shows no benefit.

Summary + source: https://offnetchat.com/news/2026-07-09/s/c99a77019ea59f71.html

#LLM #TechNews