Supported Data Formats
Tabular data files become queryable databases the moment you upload them. Zark AI supports the following data formats:| Format | File Extensions | Description |
|---|---|---|
| CSV/TSV | .csv, .tsv | Comma-separated or tab-separated values files. Simple, widely-used format for tabular data. |
| Excel | .xlsx, .xls | Microsoft Excel workbooks in both modern (.xlsx) and legacy (.xls) formats. Supports multiple sheets. Zark AI intelligently determines how to handle the structure, combining related sheets or processing as text when appropriate. |
| JSON | .json | JavaScript Object Notation files containing structured data. Ideal for nested or hierarchical data structures. |
| Parquet | .parquet | Columnar storage format commonly used in data engineering workflows. Optimized for analytical queries on large datasets. |
What You Can Ask
Once you upload a spreadsheet or data file, Zark AI automatically understands the structure and lets you ask questions like:- “How many rows are in this data?”
- “What’s the average order value by region?”
- “Show me the top 10 customers by total spend”
- “Which products had declining sales month over month?”
- “Compare Q3 performance to Q2”
Handling Messy Data
Real-world data is rarely perfect. Zark AI handles common issues automatically, including files with inconsistent formatting, European number formats (commas as decimal separators), currency symbols mixed into numeric columns, and files where the header row isn’t in the expected position.Working with Multiple Sources
Zark AI becomes particularly powerful when you combine information from multiple sources.Cross-File Analysis
Upload multiple related files and ask questions that draw from all of them. “How do our actual Q3 results compare to the projections in the plan?” uses both your results data and your planning document. “Find any inconsistencies between the contract and the proposal” compares two documents. Learn more about cross-document analysis. This works across file types too. “Based on the customer feedback in the survey responses and the metrics in the sales data, which product improvements would have the most impact?” synthesizes structured data with document content.Historical Comparisons
When you’ve built up a library of files over time, historical analysis becomes possible. Upload this quarter’s data alongside previous quarters and ask about trends, changes, or anomalies. “How has our customer retention changed over the past four quarters?” or “What’s different about this month’s results compared to our historical average?”Zark evaluates metrics relative to your historical data, helping you identify anomalies instead of relying on raw thresholds.