# Data Agent

The authors built Argo-Bench: a 2024 New York City food-delivery platform simulated into an Oracle E-Business Suite-style warehouse of 235 tables and 7.49 billion rows, where agents are graded on the actions they file — banning accounts, allocating incentive budgets, filing forecasts — against the simulator's withheld ground truth; among 14 frontier and open-weight models, the strongest, Claude Opus 5.5, solves 34.8% of tasks and averages 59.5 points.

The authors built Argo-Bench: a 2024 New York City food-delivery platform simulated into an Oracle E-Business Suite-style warehouse of 235 tables and 7.49 billion rows, where agents are graded on the actions they file — banning accounts, allocating incentive budgets, filing forecasts — against the simulator's withheld ground truth; among 14 frontier and open-weight models, the strongest, Claude Opus 5.5, solves 34.8% of tasks and averages 59.5 points.

The work introduces EvoOntology, which encapsulates an ontology as an MCP server with schema, content, and tool layers, builds an initial ontology via a builder agent issuing probe queries, and continuously refines it through attribution-guided typed edits admitted only after backbone-conditional paired evaluation, consistently outperforming ReAct baselines and static semantic-layer baselines on three data-agent benchmarks with six LLM backbones.

The work introduces EvoOntology, which encapsulates an ontology as an MCP server with schema, content, and tool layers, builds an initial ontology via a builder agent issuing probe queries, and continuously refines it through attribution-guided typed edits admitted only after backbone-conditional paired evaluation, consistently outperforming ReAct baselines and static semantic-layer baselines on three data-agent benchmarks with six LLM backbones.