Saturday, October 10, 2026

Tools

PlanetScale MCP server adds branch metrics retrieval

A new tool surfaces CPU, memory, disk, replication lag and connection readings in an AI assistant.

• MCP server: Branch metrics

  planetscale.comOct 10, 2026, 2:04 PMUTC

Agents

Retell adds evolving contact memory for voice agents

The feature creates a running brief from completed calls and SMS chats, and can use it to shape later conversations and conversation flows.

• Contact Sales Login Contact Sales Try For Free

  retellai.comOct 10, 2026, 1:07 PMUTC

Agents

Exa introduces ATLAS benchmark for search agents

The benchmark scores agents on 547 wide-search tasks, with a focus on both the correctness and completeness of results.

• Exa Site Index

  exa.aiOct 10, 2026, 1:04 PMUTC

Research

Snorkel raises Open Benchmarks Grants commitment to $30 million

The expanded program adds a benchmark red-team initiative alongside grants and a research fellowship.

• Frontier AI is Accelerating. Open Benchmarks Need to Keep Up.

  snorkel.aiOct 10, 2026, 1:04 PMUTC

Tools

Kiro CLI 2.29.0 adds configurable memory for local V3 sessions

The update lets Kiro carry useful project context between chats, with controls for memory access and automatic updates.

• October 9, 2026

  kiro.devOct 10, 2026, 1:04 PMUTC

Business

Andreessen Horowitz says it is leading investment in TypeSafe AI

The firm cites Jev’s typed outputs and reported adoption as it backs the AI company.

• Investing in TypeSafe AI

  a16z.comOct 10, 2026, 12:05 PMUTC

Infrastructure

KONST raises $30 million to expand AI data-center capacity in Asia

The Taiwan-based AI compute operator says the Series B will support regional infrastructure growth and its Token Factory strategy for enterprise customers.

• Continue Reading

  prnewswire.comOct 10, 2026, 11:04 AMUTC

Tools

Ai2 reports shorter debugging queues

Illustration

The engineering report flags a smaller baseline sample for the comparison.

• Ai2 scheduling report

  allenai.orgOct 10, 2026, 9:53 AMUTC

• Slurm Fair Tree

  slurm.schedmd.comOct 10, 2026, 9:53 AMUTC

• Hadoop Fair Scheduler

  hadoop.apache.orgOct 10, 2026, 9:53 AMUTC

Tools

Databricks adds request-tag checks and OpenJev evaluation to Unity Gateway beta

Illustration

Two Unity Gateway beta additions let custom service policies check caller-supplied tags and use OpenJev for content classification.

• October 2026 release notes

  docs.databricks.comOct 10, 2026, 9:26 AMUTC

• Service policy function reference

  docs.databricks.comOct 10, 2026, 9:26 AMUTC

• Service policy examples

  docs.databricks.comOct 10, 2026, 9:26 AMUTC

Tools

Deno joins Cloudflare, with Deno Deploy set to shut down in six months

Illustration

Deno will shift development toward a shared platform with Cloudflare, while its runtime and hosting service enter a wind-down period.

• Deno is joining Cloudflare

  deno.comOct 10, 2026, 9:07 AMUTC

• Deno is joining Cloudflare

  blog.cloudflare.comOct 10, 2026, 9:07 AMUTC

• Cloudflare Durable Objects

  developers.cloudflare.comOct 10, 2026, 9:07 AMUTC

Agents

Claude Managed Agents adds beta dynamic workflows

The feature lets agents run phased, multi-agent workflows in the background, with setup and run tracking through the session event stream.

• anthropic

  platform.claude.comOct 10, 2026, 8:03 AMUTC

Research

OpenProblemBench evaluates AI on 82 unresolved science problems

The benchmark covers open questions in mathematics and theoretical physics, with evaluators judging submissions for correctness, completeness and progress.

• OpenProblemBench: Benchmarking AI on Open Problems in the Foundational Theoretical Sciences

  arxiv.orgOct 10, 2026, 12:16 AMUTC

Research

AgentHorizon tests AI judges on long computer-use tasks

The benchmark pairs computer-use trajectories with closely related instructions to measure whether judges can spot incomplete or incompatible task outcomes.

• AgentHorizon: Evaluating Agentic Judges for Long-Horizon Computer-Use Tasks

  arxiv.orgOct 10, 2026, 12:05 AMUTC