Agent control over candidate processing improves search agent performance

Programmatic Search Agent outperforms query-only agents by extending agentic search to processing and presenting retrieved evidence

Chinese Tech
Jiaming Qian · Huiyan Yang · Mandi Liu · Jie Liu · Wenkai Shen · Pengyang Zhou · +4 more

Zhejiang University · Yuanbao Team, Tencent · Peking University

Research Digest··3 min read
A team from Zhejiang University, Tencent, and Peking University introduces Programmatic Search Agent (PSA), which gives agents control over how retrieved candidates are processed and presented, not just query reformulation.

The authors first analyzed trajectories of standard query-based agents on the BrowseComp-Plus benchmark, finding that supporting passages can be retrieved but never appear in the agent's observations.

Why this paper

From Yuanbao Team, Tencent and 2 others

In one line

Programmatic Search Agents improve search by letting agents control how retrieved candidates are processed and presented.

What we could check

  • ·No code link found
  • ·No weights link found
  • ·No dataset link found
  • ·No compute details found
  • ·No stated limitations found
  • ✓Reports numbers on named benchmarks (2 benchmarks)

Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.

§

Research Digest

Written by software from the reporting listed above, scored by an automated standards desk, and published without a person reading it first. If something here is wrong, tell the editor and it will be put right.