Topic Briefing

Artificial Intelligence

Executive Summary

Anthropic and OpenAI are shipping autonomous agents that build entire compilers and ship tens of thousands of lines of code in days. While tech firms like Microsoft block billions in AI-aided scams, local communities are blocking the data centers needed to run these models as residential electricity prices surge. In healthcare, California is scaling AI data unification for 2 million patients to automate risk quality and care coordination. Key Takeaways: • Orchestration beats manual coding: OpenAI Lead Sherwin Wu reports 95% of their engineers now manage 10-20 parallel agents, cutting code review times from 15 to 3 minutes. • Washington state legislation proposes a "BYONCE" (bring your own new clean energy) model, requiring data center developers to fund their own grid connections to protect residential ratepayers from infrastructure costs. • A UCSF study of 310 physicians found that AI scribes primarily benefit slow documenters (10% time reduction), while efficient documenters see negligible gains, suggesting health systems should target high-burden clinicians for maximum ROI.

Key Themes

Major trends and developments identified from this week's coverage

Agentic Software Engineering

Anthropic and OpenAI are deploying autonomous agents that manage complex systems-level tasks like writing CUDA kernels and building C compilers, shifting the developer's role from writing to orchestrating.

3 articles

Data Center Infrastructure Backlash

Local governments in Michigan, Indiana, and Washington are enacting moratoriums on data centers as rising electricity costs—up to 267% in some wholesale markets—impact residential rates.

4 articles

Healthcare AI Scale

Major health systems and state programs are moving from AI pilots to industrial-scale deployments, with California implementing a statewide platform for 2 million Medi-Cal patients and UCSF physicians saving 1.5 hours weekly via AI scribes.

3 articles

AI-Enhanced Security Operations

Microsoft and security analysts are using AI models to block $4 billion in scams and map Russian APT28 tactics, while open-source projects launch tools to detect rogue local AI agents like OpenClaw.

2 articles

Key Players

Top companies, people, and technologies mentioned this week

OpenAI
Company●●●●●

32 articles

Developed the GPT-5.3 Codex model used in the head-to-head comparison. (+30 more)

Anthropic
Company●●●●●

23 articles

Developed the Claude Opus 4.6 and Opus 4.6 Fast models used in the comparison. (+21 more)

Google
Company●●●●●

21 articles

Major technology company and former employer of the author entering a partnership. (+19 more)

Simon Willison
Person●●●●●

21 articles

Author of the notes and post (+20 more)

Microsoft
Company●●●●●

8 articles

Reportedly encouraging employees to use multiple AI tools, leading to the rapid dominance of Claude Code within its engineering teams. (+7 more)

Meta
Company●●●●●

11 articles

One of the major tech companies that has begun allowing candidates to use AI assistants in technical sessions. (+8 more)

NVIDIA
Company●●●●●

5 articles

Manufacturer of the H100, A100, and T4 GPUs for which the skill provides optimization guidance. (+3 more)

Innovaccer Inc.
Company●●●●●

4 articles

Collaborates with American Medical Group Association (AMGA), Collaborates with Colorado Access

GitHub
Company●●●●●

6 articles

Platform used to manage the 44 PRs and 98 commits generated during the sprint. (+5 more)

Claude Code
Technology●●●●●

12 articles

AI tool mentioned as using subagents in parallel (+11 more)

Editor's Picks

AI-recommended articles based on relevance and quality

Top Sources

Want personalized briefings?

Kelp creates AI-powered briefings tailored to your specific interests and sources. Get the insights that matter most to you.

Start Your Free Briefing