An agentic AI system that lets users query Excel and CSV files in natural language — analyzing data, charting, flagging quality issues, and explaining results.
Business Problem
Non-technical users regularly need to analyze spreadsheets but can't easily write SQL, build visualizations, check data quality, or interpret raw analytical output.
Users
Business users working with Excel and CSV data who need answers, charts and reports without a spreadsheet-analysis background.
Constraints
Solution
I built a LangGraph multi-node pipeline that classifies each request and routes it to a specialized path — SQL analysis over DuckDB, visualization, report generation, data-quality checks, semantic search, or a natural-language explanation of the result.
Architecture
Your Contribution
Technology Stack
Results
Challenges & Trade-Offs
Flexible agents vs. deterministic workflows
Conditional routing lets the system pick the right specialized path per request instead of forcing one rigid pipeline — trading some predictability for coverage across very different analytical tasks.