Search papers, labs, and topics across Lattice.
Waseda University Tokyo
2
0
5
Judging AI coding agents solely by merged/rejected pull requests is misleading: over 60% of rejections aren't the agent's fault, and even "successful" merges often hide significant human intervention.
Forget monolithic LLMs for requirements engineering: QUARE's multi-agent negotiation framework achieves 105% better compliance by explicitly surfacing and resolving inter-quality conflicts.
