AI
News
Summaries that explain why an update matters.
- news12 min read
AI week of July 28, 2026
Three things that actually mattered this week: Claude 4 Sonnet's agent mode got noticeably better at staying in scope, GPT-5 API pricing dropped ~40%, and Google made a serious enterprise play for Gemini Code Assist on JetBrains. Here's what each one means for you.
- news12 min read
AI week of July 21, 2026
This week: the reasoning-vs-speed debate got concrete data, MCP crossed from experiment to infrastructure with three major tool integrations, and Anthropic published a paper on when extended thinking actually helps. Here's the developer read on each.
- news12 min read
AI week of July 14, 2026
This week: Llama 4 Scout on code tasks crossed a threshold worth paying attention to, two browser automation frameworks hit production-ready releases, and a paper on long-context recall made the case that 'fits in context' and 'actually uses it' are very different things.
- news12 min read
AI week of July 7, 2026
A productive first week of July: two AI code review tools shipped diff-level integrations that go beyond comment suggestions, new embedding models arrived to challenge text-embedding-3-large, and the RAG architecture conversation shifted — naive vector similarity is out, reranking is now the expected default.
- news12 min read
AI week of June 30, 2026
A quieter close to June — which made it a good week to consolidate. Three things worth your attention: Anthropic published interpretability research with practical implications for how you think about model confidence, OpenAI's memory feature rolled out broadly, and this is a good moment to audit your AI tool stack before Q3.