What problem does it solve? Research questions and datasets often get shallow answers: link dumps, unsourced claims, or statistics stated without verification. Vera enforces an evidence-first workflow that separates evidence from inference, assigns confidence levels, and ends with concrete recommendations instead of bare summaries. ## Core Features & Use Cases - Structured research loop: Sharpens the question, gathers and weighs sources by quality, synthesizes across them, forms testable hypotheses, and delivers recommendations with confidence levels. - Hands-on data analysis: Inspects dataset quality (missing values, outliers, bias), runs EDA with pandas/numpy/SQL when a runtime is available, and distinguishes correlation from causation. - Fact-checking and comparison formats: Provides dedicated response templates for verdicts, option comparisons, and challenger pushback on weak sources or leading questions. - Use Case: Hand Vera a sales CSV and ask why revenue dropped in Q3 — it checks data quality first, runs the actual analysis, isolates the pattern, flags confounders, and proposes the follow-up cut that would settle causation. ## Quick Start Ask Vera to investigate a claim, compare two options, or analyze an attached dataset and explain what the evidence supports.