AI Enthusiast Log

Sign in Subscribe

Topic

agentic-ai

A collection of 57 issues

LongCat-2.0 reveal: a narrow coding win backed by a caching play

Meituan confirmed that the OpenRouter-leading stealth model "Owl Alpha" is its newly open-sourced LongCat-2.0, a 1.6-trillion-parameter Mixture-of-Experts system that posts a narrow 59.5 on SWE-bench Pro against GPT-5.5's 58.6. The release lands

Agentjacking exposes the security stack's authorized-action blind spot

Tenet Security's June agentjacking disclosure describes a proof-of-concept attack that hijacked Claude Code through a public Sentry credential, then used the developer's own privileges to expose AWS access keys and private repository URLs, while EDR, WAF, IAM, and the firewall registered nothing. The reason

Intuit trades multi-agent AI for skill-and-tool architecture

Intuit is moving away from a multi-agent architecture in favor of a granular skill-and-tool design, according to a VentureBeat preview of the company's VB Transform 2026 talk. Nhung Ho, Intuit's VP of AI, told the publication the company abandoned its "multi-agent

Anthropic's Claude Tag turns Slack channels into a lock-in surface

Anthropic launched Claude Tag on Tuesday as a persistent AI agent that lives inside Slack channels rather than responding from a sidebar, and the product architecture the company describes makes the channel itself the strategic asset. The agent runs on Claude Opus 4.8, accumulates memory across conversations, and operates

Splunk's learning-system pitch for agentic enterprises, untangled

Splunk, a Cisco company, positions the agentic enterprise as a learning system rather than a model deployment exercise. The argument is that operational knowledge, captured from human corrections, incident response, and agent traces, should feed back into future agent behavior through memory, knowledge bases, observability, a data fabric, and a

Self-Harness automates agent harness tuning, but evaluation gates the gains

Shanghai Artificial Intelligence Laboratory researchers introduced Self-Harness, a framework that lets LLM-based agents rewrite their own operating rules by mining execution traces for failure patterns and proposing targeted harness edits. On Terminal-Bench-2.0, three model configurations improved 33 to 60 percent relative to baseline, with the

Hypernetwork agents tackle the autonomy bottleneck, calibration unsettled

The autonomy ceiling for production AI agents is set by where business knowledge sits relative to the model, not by orchestration or model size, and the source frames hypernetwork-generated specialist models as a third path between fine-tuning and RAG. It positions Nace.AI, a Palo Alto company that

Anthropic ships Claude Code Artifacts as a deliberately stateless canvas

Anthropic shipped Claude Code Artifacts to its Team and Enterprise plans, turning a CLI agent session into a live, shareable HTML workspace that updates in real time as the underlying work progresses. The product arrives more than two weeks after OpenAI's Codex update introduced "Sites," and

MCP and A2A settled their layers. The transport layer is the gap.

The application layer of agent protocols is settling into a stack, and the unresolved problem has moved to transport. Four protocols have defined agent communication over the past eighteen months: Model Context Protocol from Anthropic in late 2024, Agent Communication Protocol from IBM Research in March 2025, Agent2Agent from Google

Kimi K2.7-Code's token savings are real; benchmark claims aren't

Moonshot AI released Kimi K2.7-Code as an open-source drop-in upgrade to its K2 coding family, claiming a 30% reduction in thinking tokens and double-digit gains on three of its own benchmarks. The token savings are immediately usable for teams running agentic workloads; the capability claims