Search as Code

Assess

Techniques

A search-oriented agent workflow that treats retrieval as a repeated, programmable part of task execution.

Why it's here

Placed in Assess: 1 article(s) of evidence from 1 source(s), led by research-stage coverage, with 1 in the last 30 days. Confidence 24%. Low accumulated evidence, so it defaults conservatively pending more signal.

Evidence (1)

7The New Stack·10/15/2026research
Agentic AI Falls Short When Retrieval Is Weak
BrowseComp-Plus results show that the same model can perform very well when supporting documents are provided directly, but accuracy drops sharply when an agent must retrieve evidence on its own using BM25. The article argues that for agentic systems, retrieval quality often matters more than adding model intelligence, especially in messy production data environments.