Report: Progressive Disclosure of Agent Skills
Workday finds progressive disclosure of agent skills improves retrieval quality but slightly increases latency.
Workday reports on lazy-loading named procedures, or skills, as its deployed LLM agents' skill libraries grow and raise operating cost. Progressive disclosure was expected to reduce cost, but its effects on latency and retrieval quality were unclear. Empirical tests find that progressive disclosure improves skill-retrieval quality while only marginally degrading overall latency.
- Named skills expand Workday LLM agents but raise operating cost.
- Progressive disclosure improves skill-retrieval quality.
- Overall latency degrades only marginally.
From the original article
1 min read at arxiv.orgUsers of Workday's deployed LLM-based agents often request features which can be addressed by defining named procedures, also known as skills, in the LLM context, effectively augmenting agents' capabilities. However, as an agent's skills library grows in size, so does the agent's operational cost.
By Guilin Zhang · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.