When vector search is weak for agent memory
works well when an agent needs to find text by meaning. It can be weaker when the memory is structured, such as which option was chosen, which account is involved, or what next action fits the other fields in a row. can blur the exact structure that matters for these lookups.
A different approach is to query the agent’s own structured history, predict the needed field, and attach a . If the is low, the system can fall back to the LLM. In one dataset, this recovered the right context about 65% of the time from a small amount of data, including cases that missed.
The limit is clear: are still better for language-heavy recall. This method is aimed at repeated, high-volume structured lookups inside an .
Key points
- can miss structured that depend on exact fields and relationships.
- are useful for meaning, but they can blur row-and-field structure.
- The proposed method predicts a field from the agent’s structured history and checks .
- Low can trigger a fallback to the LLM instead of using it every time.
- The reported result was about 65% right-context recovery on one dataset with little data.