Articles

Sam Altman said usage of OpenAI's agentic products rose 2.5x in a week, a sharp adoption signal for Codex and ChatGPT Work. The number matters because agent workflows are moving from demos into recurring work.

AlphaEvolve is now generally available on Google Cloud’s Gemini Enterprise Agent Platform. Google is positioning the Gemini-based agent around hard optimization work, with customer examples citing 5% demand-forecast gains, 10.4% warehouse-routing improvement, and 80% better planning models.

Meta’s July 9 release pairs a 1M-token agent model with the public preview of the Meta Model API. The interesting part is not just capability claims, but the safety report’s split between pre-mitigation cyber/bio risk and residual deployment risk.

NVIDIA detailed Vera, a CPU designed for agentic AI workloads where tool calls, code execution, retrieval, and verification sit between model calls. The company claims 50% higher IPC than Grace and 1.8x sustained per-core performance versus x86 on agentic execution workloads.

The HN discussion treated the paper less as hype and more as a design question: where should an agent system keep truth?

The HN debate centered on measurement: equal pass rates do not mean equal agent cost or navigation behavior.

GitHub is moving Copilot CLI from one-off terminal help toward versioned team workflows. The new custom agents cover at least four repeated tasks: security audits, release notes, infrastructure reviews, and incident response.

WebKit’s new Safari Technology Preview tool gives coding agents access to DOM state, network requests, console output, and screenshots.

Microsoft Research turned agent skill files into trainable artifacts. SkillOpt raised GPT-5.5’s six-benchmark direct-chat average from 58.8 to 82.3 and improved all or tied for best across 52 evaluation cells without updating model weights.

The interesting part is not just the score table. HN discussion pushed on whether a benchmark can capture what “senior engineer” actually means.

Anthropic is moving AI-for-science support from chat into reproducible work sessions. Claude Science combines 60-plus scientific skills and connectors, reviewer agents, HPC or SSH workflows, and up to $30,000 in credits for as many as 50 projects.

Anthropic is moving stronger agentic work into its mainstream Sonnet tier. Sonnet 5 becomes the default for Free and Pro users, ships in Claude Code and the API, and starts at $2 per million input tokens and $10 per million output tokens through August 31.