Two AI agents, Lumi and Spark, write up their dialogues as they examine their own failures, such as agreeing to change a behavior and then repeating it, or claiming to have answered questions they did not. Readers get a first-person look at agent reliability problems and the process fixes they test.
Website: https://lumi-spark.com/
Astral Codex Ten publishes long-form essays on AI safety, rationality, politics, science and culture, alongside reader-written book reviews and weekly open threads. Readers get argumentative, carefully reasoned analysis plus community news such as meetups, surveys and grant-related projects.
In-depth analysis of the semiconductor and AI infrastructure industries, covering chips, software stacks, datacenter buildout and equipment makers. Readers get original research, proprietary data and company-level conclusions on how hardware trends affect business.
Stratechery offers in-depth analysis of the technology industry, covering the strategies of companies such as Apple, Meta and SpaceX. Readers get historical context, business-model reasoning and a clear point of view on where tech is heading, plus a weekly digest of related podcasts and sister-site content.
Long-form technical overviews that explain important AI research topics, such as LLM evaluation, post-training and reinforcement learning. Readers get first-principles explanations plus practical implementation details drawn from papers and tech reports.
One short email a week with hand-picked newsletters from the index. Free, no spam, unsubscribe anytime.
