Apple: AI agents should plan first only sometimes
Why it matters
Apple researchers' paper "Plan or Retrieve First? A Scale Study of Tool Discovery Architectures" (arXiv, Oct 13; accepted at EMNLP 2026 Industry Track) compares search-first (embed the request, retrieve similar tools) with plan-first (the model drafts hypothetical tool stubs, then searches with them) on ToolBench's 46,984 APIs with Gemini 2.5 Flash. Single-tool requests: search-first recalled 37.2% of the right tools in its top 5 vs 28.9%. Multi-tool requests within one category: plan-first 18.8% vs 14.3%; across categories the gap was not significant. Plan-first adds ~75-104% token overhead plus an extra model round trip; routing by task type cut ~36% of that overhead with no retrieval loss, though a real router is untested. Planners used the retrieved tools at least 96% of the time, so retrieval decides what an agent can do.
Read the source: arXiv (Apple) (opens in a new tab)
All hits from 18 Oct 2026More AgentsGet the daily hits by email