Workday Forms AI Research Team for Enterprise AI

Workday, Inc. has introduced Workday AI Research, a technical research team focused on reliable, trustworthy, and efficient AI for enterprise software, the company announced in a press release. The team studies areas including agent memory, explainability, orchestration across multiple agents, reward overoptimization, recommendation systems, and adaptive resource control.

Workday said its researchers have had recent work accepted by the International Conference on Machine Learning, the International Conference on Learning Representations, the ACM Web Conference, and the Association for Computational Linguistics.

The company highlighted research on selective agent memory that raised precision by 12 percent, improved memory quality by about 8 percent, retained 97 percent of useful memories, and ran about 31 percent faster than the comparison system. Another study found that assigning specialized agents to explore options, check rules, and coordinate work improved accuracy by 5.8 percent, with all final answers meeting the study constraints.

In separate work, researchers found that deleting information from an AI agent memory did not always remove copies from older summaries. When asked to forget information, a copy remained recoverable from an old summary about one in five times.

Workday AI Research will also run the Workday AI Research PhD Fellowship. Fellows receive $50,000 in annual research funding through an unrestricted gift to their university, mentorship from a Workday AI researcher, direct collaboration access, and early access to relevant career opportunities.