Support
Research Agent
Tool orchestration, sessions, evidence handling, and AI interpretation limits.
Validation disclosure. Platform inventory reports independentlyBenchmarkedCount = 15. 15 estimator inventory entries have passed independent numerical benchmark comparisons for specified test configurations. These results do not establish complete econometric validation of every model, assumption, inference procedure, or supported input. Internal tests and native in-app EXACT MATCH for three paths are not independent package benchmarks.
Topics in Research Agent
All categories- Research Agent overview
Bounded agent loop that coordinates existing IES tools. It does not invent instruments, datasets, or statistical values.
- Research Agent limitations
AI-generated interpretations require researcher review. Tools are capped; planned engines stay unavailable.