Cloudflare claims Clairic model achieves 49% cost reduction in AI pipeline testing

Meta engineers reduced AI hallucinations and data prep time by splitting a single AI agent into 14 specialist agents that handle different tasks.

Google Gemini can now assemble outfits from item screenshots without needing product listings, raising questions about whether retailers will know when their goods are recommended.

A Basis survey of 184 marketers shows ChatGPT adoption declining to 74.5%, Claude rising to 72.3%, and 51% now using AI agents—yet only 6.5% measure AI ROI against benchmarks.

Meta separated AI training permissions from its commercial crawling terms, allowing crawl access to persist while restricting model training use.