📊 ADA turns CSV and Excel files into business dashboards without configuring a BI platform. It cleans data, detects metrics and segments, identifies anomalies, calculates a guarded baseline forecast, and recommends what to investigate next.
Its main difference is traceability. Natural-language questions become typed query plans executed locally with pandas. Every answer shows the calculation behind it, while interpretations are clearly separated from observed facts.
🤖 AI is optional. When enabled, the model receives the schema and calculated evidence, never the original rows or cell values. A plan that cannot be resolved is rejected instead of producing a guess.
The author built an open-source Streamlit application combining deterministic analysis, Plotly charts, anomaly detection, forecasting, and an auditable conversational layer.
💡 Explanation in a nutshell#
ADA works like an analyst who performs the calculations first and then explains the results. AI helps formulate questions or summarize findings, but the source of truth remains reproducible calculations over local data.
More information at the link 👇
