Research watch
Curated stories prioritize research implications over volume. Official sources and independent reporting remain visibly distinct.
Curated major stories
Each record includes the development and a short explanation of why it matters to this research landscape.
Benchmark · Mercor / evaluation
Mercor's APEX program uses expert-authored tasks to examine professional work in areas such as medicine, law, finance, and consulting.
Evaluation is shifting from exam-style questions toward realistic deliverables and domain-specific judgment.
Company development · AfterQuery / agent training
AfterQuery published work connecting expert-curated trajectories and tooling with improved results on an agent-oriented terminal benchmark.
Training-data providers increasingly combine data production with applied research and evaluation design.
Research direction · Fleet / environments
Fleet describes simulated worlds and real-world challenges designed to model work for the training and evaluation of AI agents.
Environment design is emerging as a distinct infrastructure layer for agent development.
External feed
A future RSS pipeline will be separated from editorial selections, retain publisher links, and fail gracefully when feeds are unavailable.