{ "schema_version": 1, "builder_plan": { "scenario": "An agent looks good in demos, but you cannot tell which product tasks it fails.", "change": "Use the study's evaluation idea to inspect a small set of tasks with human-checked answers.", "baseline": "Your current evaluator and a manually reviewed reference set.", "steps": [ "Locate the paper's implementation and check its license, model support, data needs, and hardware requirements. Stop here if you cannot obtain the required artifacts.", "Save a small set of representative product tasks and expected outcomes. Keep evaluation tasks out of any training or tuning, and choose quality and cost limits before running.", "Run the baseline and the proposed change on identical inputs. Keep other settings fixed and repeat paired runs to expose variation.", "Save per-task outputs, failures, timings, and all usage costs. Review regressions before expanding the test." ], "measure": "Agreement with human review, missed failures, false alarms, and evaluation cost per task.", "decision": "Keep the change only if it meets your preselected quality and latency limits and improves the metric you care about after setup and operating costs. Otherwise retain the baseline. This is an exploration plan, not a validated recommendation." }, "builder_review": null, "arxiv_id": "2609.25192", "title": "FinFIRST: Benchmarking Search Agents for Financial Information Retrieval, Sourcing and Traceability", "summary": "FinFIRST replaces final-answer-only evaluation of financial search agents with atomic rubrics that separately score raw-information acquisition, source verification, and computation and answer formation, making pipeline errors localizable. It costs evaluation effort rather than serving resources: adopting it means building or reusing its rubric decomposition and reference packages, and running 123 expert tasks against your system via API or local models. It can fail as a proxy if your production tasks diverge from its expert-authored distribution, if rubric grading itself is noisy, or if a high atomic score masks poor latency and cost behavior that the benchmark does not measure.", "rationale": "For a constrained team shipping retrieval-augmented financial agents, process-level rubric evaluation is directly usable through third-party APIs and gives diagnostic value beyond answer accuracy, but its transfer to the reader's specific workload must be validated before it gates releases.", "technique": "Evidence-traced rubric benchmark for financial search agents", "family": "avaliacao_de_agente", "recommendation": "testar", "published": "2026-09-21", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2609.25192", "independent_implementations": 0, "total_implementations": 0, "stars": 0, "citations": 0, "claims": [ { "claim_id": "claim-01", "statement": "FinFIRST replaces final-answer-only evaluation of financial search agents with atomic rubrics that separately score raw-information acquisition, source verification, and computation and answer formation, making pipeline errors localizable. It costs evaluation effort rather than serving resources: adopting it means building or reusing its rubric decomposition and reference packages, and running 123 expert tasks against your system via API or local models. It can fail as a proxy if your production tasks diverge from its expert-authored distribution, if rubric grading itself is noisy, or if a high atomic score masks poor latency and cost behavior that the benchmark does not measure.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.25192", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "Best reported results on the benchmark: Claude-Opus-5 at 87.59% atomic score and GPT-5.6-Sol at 71.54% strict pass rate, with computation and answer formation lagging raw-information acquisition across all 15 configurations.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2609.25192", "source_page": null, "source_excerpt": "" } ], "exposure_map": [ { "dimension": "quality", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "compute", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "latency", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "operations", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "compatibility", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "security", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "data_and_training", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "reproducibility", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" } ], "risks": [], "minimum_test": [], "open_questions": [], "independent_tests": [], "lab_reproductions": [], "report_available": false, "validation_tier": "unknown", "evidence_tier": "unknown", "infrastructure_basis": "unknown", "equations": [], "equations_status": "not_fetched", "core_kind": "", "equations_fetched_at": "", "equations_source_url": "", "equations_source_sha256": "", "exposures_reviewed_at": "", "exposures_source_sha256": "" }