Workday Launches AI Research Team to Enhance Enterprise AI Reliability and Efficiency

Workday AI Research: Advancing Reliability and Efficiency in Enterprise AI

Workday, a leader in enterprise cloud applications for finance, HR, and planning, has announced the formation of a dedicated AI research team known as Workday AI Research. This initiative is aimed at enhancing the reliability and efficiency of enterprise AI, addressing the complex technical challenges that come with it.

As organizations increasingly integrate AI into their operations, the need for powerful and trustworthy AI systems has never been higher. Workday AI Research is not solely focused on developing new AI technologies for Workday’s applications; it aims to provide valuable insights and methodologies to the broader AI research community as well.

The team at Workday is investigating various crucial factors in enterprise AI, such as long-term memory capabilities of AI agents, explanations for AI decisions, multi-agent orchestration, potential over-optimization in AI learning, recommendation systems, and adaptive resource management. The outcomes of some of their investigations have been presented at major international conferences including the International Conference on Machine Learning (ICML) and the Association for Computational Linguistics (ACL).

Supporting the Next Generation of AI Researchers

In addition to these research efforts, Workday has initiated the "Workday AI Research PhD Fellowship". This fellowship is designed to support outstanding PhD students who are conducting research in the domains of AI and enterprise software. Selected recipients will benefit from annual funding of $50,000 for unrestricted research purposes at their institutions, as well as professional mentoring from Workday AI researchers and opportunities for direct collaboration. This initiative marks a significant commitment by Workday to nurturing the next generation of AI experts.

Latest Research Findings

The Workday AI Research team is delving into some pressing inquiries that businesses face when implementing AI:

1. Selective Memory for AI Agents: Can AI agents retain only the necessary information instead of storing everything? Workday’s researchers have developed a selective memory approach that enables AI agents to function better by maintaining useful data while discarding outdated, redundant, or unreliable information. This strategy has shown a 12% increase in precision and an 8% enhancement in overall memory quality. Notably, while maintaining 97% of essential information, the processing was approximately 31% faster compared to leading AI systems.

2. Improved Decision-making through Multiple AI Agents: Can multiple AI agents make better decisions than a single agent? The research indicates that by dividing tasks among specialized agents—one to generate options, another to ensure compliance with rules, and a coordinating agent—decision quality and adherence to protocols improve. Results showed a 5.8% improvement in accuracy while satisfying the defined constraints.

3. The Challenge of 'Forgetting' in AI Memory: When directed to forget certain information, do AI agents succeed in fully deleting it? The findings revealed that even when instructed to forget, the information could still be recoverable in about 20% of instances, indicating that complete deletion requires removing all references to that information.

As enterprises navigate the rapid changes brought about by AI, there is a clear need for systems that not only retain relevant context but also provide trustworthy recommendations with stringent compliance frameworks. Workday AI Research aims to connect rigorous AI research processes with the practical needs of businesses, developing next-generation AI tools that enhance usability, efficacy, and transparency in real-world applications.

Gerrit Kazmaier, President of Product and Technology at Workday, emphasized, “As AI agents begin to understand context and execute functions on behalf of employees, organizations will face intricate challenges related to privacy, auditability, efficiency, and enterprise-grade accuracy. Workday AI Research is dedicated to addressing these challenges, providing the scientific insights necessary to develop intelligent, trustworthy AI systems that organizations can confidently adopt at scale.”

For additional insights and updates on Workday AI Research, visit workday.com/ai-research.

About Workday

Workday provides enterprise cloud applications for finance, HR, and IT that are built on a foundation of reliability and accuracy. Through the integration of AI into core business functions, Workday allows organizations to leverage autonomous AI agents to efficiently perform tasks and generate measurable outcomes. With over 11,500 customers, including more than 65% of Fortune 500 companies, Workday exemplifies a modern solution for enterprise AI needs.