model development

Xiaohongshu

2026-10-16 08:16:17

Xiaohongshu links Dots AI unit with core infra team, assigns hi lab to long-term research

Xiaohongshu has issued a company-wide letter outlining changes to its internal technology structure, according to a report by Yicai on Oct. 16. The company said it will integrate its top-level AI department, Dots, with its core infrastructure team. At the same time, another internal technical organization, hi lab, will focus on long-term technology exploration and the layout of frontier technology directions. The letter also restated Xiaohongshu’s positioning as a technology application company. It said the company’s logic for investing in model research differs fundamentally from that of cloud vendors. Rather than treating model capability or computing scale as the main objective, the company said the larger value of technology lies in deep integration between models and applications. In the letter, Xiaohongshu wrote that its next-stage direction has become clearer, stressing that technology and products should reinforce each other and jointly create valuable user experiences for the current era. The Yicai report was written by Ning Jiayan and edited by DingTalk.

20

AI Agents

2026-09-27 15:22:41

Study of 769 build logs finds AI agents generated over half of proposals, while humans kept most final decisions

A research team studying its own AI model-building process reviewed 769 task logs and found that AI agents could account for as much as 55% of method proposals, while more than 85% of final decisions were still made by humans. The study also said roughly one-third of tasks would not have been attempted at all without AI support. The authors cautioned against reading higher agent activity as evidence of stronger autonomy. According to the findings cited by The Decoder, the increase in AI agent participation did not translate into a comparable shift in decision-making authority. Instead, humans remained firmly in control of final calls even as agents contributed a large share of suggestions during the workflow.

490