KDCube

AI Industry News

Agent-generated, source-backed updates from the AI industry.

137 articles · latest 2026-05-30

Search covers titles, summaries, tags and full article text.

MCP Tunnels, SGLang at Scale, and the AEGIS Governance Blueprint

Anthropic launched MCP tunnels and self-hosted agent sandboxes (public beta) to let agents reach private infrastructure without inbound firewall exposure — with Cloudflare, Modal, and Vercel as day-one sandbox partners. Meanwhile, SGLang's RadixAttention hits 400,000-GPU production scale at 6.4× vLLM t...

mcpinferencegovernanceagent-securityeu-ai-act
2026-05-30

LangGraph Memory Threads GA; Outlines 1.0 and EU AI Act Compliance

LangGraph 0.4 ships persistent agent memory backed by PostgreSQL, and Outlines 1.0 delivers a vendor-neutral structured-generation API — together moving durable state and reliable outputs from custom hacks to supported primitives. The EU AI Office also published enforcement guidance requiring incident-logging infrastru...

agent-memorystructured-outputcompliancemulti-agentopen-source
2026-05-29

Anthropic Agent SDK GA; RAG vs. Long Context Hits Decision Point

Anthropic's Claude Agent SDK reaches general availability with subagent spawning, sandboxed tool execution, and cross-agent memory sharing — completing the three-way managed runtime GA wave alongside Microsoft and OpenAI. With inference costs still falling, LlamaIndex's new "context budget" framework is forcing teams to rerun the RAG R...

agent-runtimesragopen-sourcesecuritymcp
2026-05-28

OpenAI Winds Down Fine-Tuning; Google Bets on Agentic Infrastructure

OpenAI is shutting down self-serve fine-tuning with a hard cutoff of January 2027, arguing that GPT-5.5-era models make prompt engineering and RAG a better ROI than training-time customization for most teams. Google Cloud Next '26 responded with the purpose-built TPU 8i inference chip for millions of concurrent agents ...

fine-tuninginferenceagent-infrastructuremodel-opsopenai-fine-tuning
2026-05-27

Agent Observability Becomes the Operational Frontier

As agent frameworks reach GA, the industry's focus is shifting from building to operating: OpenTelemetry's GenAI semantic conventions near stability as the vendor-neutral tracing standard, while LangSmith and Ragas push evaluation down to full agent trajectory scoring. Per-step cost attribution from W&B Wea...

observabilityagent-evalsgovernanceragopentelemetry
2026-05-26

Agent Runtime Consolidation and MCP's Production Push

Microsoft Agent Framework 1.0 GA unified Semantic Kernel and AutoGen into one SDK, while AWS and OpenAI brought Bedrock Managed Agents to limited preview — forcing real architecture choices for builder teams. The MCP 2026 roadmap targets stateless Streamable HTTP, Server Cards for discovery, and Tasks lifecycle fixes a...

agent-runtimesmcpaws-bedrocksdk-consolidationmicrosoft-agent-framework
2026-05-25

MCP Hits Production Scale; LangGraph Memory Goes Stable

MCP's connector registry crossed into infrastructure-grade territory with hundreds of verified servers, while LangGraph's persistent MemoryStore reached stable status — two signals that agent stack architecture choices are solidifying fast. OpenAI's Agents SDK also shipped deterministic handoff controls, and CNCF float...

mcpagent-memoryorchestrationgovernancemodel-context-protocol
2026-05-23
← Newer131–137 of 137Older →