{ "schema_version": 1, "builder_plan": { "scenario": "An assistant chooses the wrong API or sends arguments your product cannot use.", "change": "Apply the proposed tool-selection or argument-generation change in a sandbox.", "baseline": "Your current agent with the same tool schemas and task budget.", "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": "Correct completed actions, invalid calls, retries, latency, and cost per completed 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": "2610.03524", "title": "From Benchmarks to Production: A Text-to-SQL System for Complex Financial Data", "summary": "FLINT replaces general-purpose Text-to-SQL prompting (Spider/BIRD-style) with a pipeline combining a lookup agent that resolves natural-language concepts to opaque integer key constraints, embedding-based retrieval over an expert-authored query template bank, and foreign-key-traversal schema linking. The cost is engineering and maintenance effort: a curated template bank, reference-table lookup infrastructure, and schema-traversal logic, all specific to the target database, plus extra LLM calls and retrieval latency per query. It can fail when the template bank lacks coverage of new question types, when the lookup agent mis-resolves a concept to wrong IDs, or when schemas drift and the expert-authored components are not updated.", "rationale": "The approach is feasible on a single GPU or API-only stack and directly addresses a real production failure mode, but its gains depend on domain-specific curated assets whose value must be validated on the reader's own schema and workload.", "technique": "Domain-specialized Text-to-SQL with lookup and schema linking", "family": "uso_de_ferramenta", "recommendation": "testar", "published": "2026-10-02", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2610.03524", "independent_implementations": 0, "total_implementations": 0, "stars": 0, "citations": null, "claims": [ { "claim_id": "claim-01", "statement": "FLINT replaces general-purpose Text-to-SQL prompting (Spider/BIRD-style) with a pipeline combining a lookup agent that resolves natural-language concepts to opaque integer key constraints, embedding-based retrieval over an expert-authored query template bank, and foreign-key-traversal schema linking. The cost is engineering and maintenance effort: a curated template bank, reference-table lookup infrastructure, and schema-traversal logic, all specific to the target database, plus extra LLM calls and retrieval latency per query. It can fail when the template bank lacks coverage of new question types, when the lookup agent mis-resolves a concept to wrong IDs, or when schemas drift and the expert-authored components are not updated.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.03524", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "FLINT outperforms state-of-the-art Text-to-SQL baselines using the same LLM on 359 questions over production financial schemas, where general-purpose systems score below 50%; no multiplicative factor is reported.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.03524", "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": "" }