SifarData is a chat-based tool for CSV and Excel files. You open a file, type a question in plain English, and get back an answer, a table or a chart. The file is never uploaded. It is read and processed in your browser, on your own computer.
Why I built it
Most AI data tools upload your file to their servers first. That breaks down in two common situations. If the file holds customer or financial data, sending it to a third party is often against company policy. If the file is large, say a 2 GB export, the upload is slow or hits a size cap before any analysis starts.
How it works
1. Open a CSV or Excel file. No account is needed to load it.
2. Ask a question (a free account is needed for this). The AI receives the column names, data types, a few sample rows and short summaries of query results, then writes a query.
3. The query runs locally against your full file. You see the result, the chart and the SQL behind it.
If you don't want any rows shared, turn on strict privacy mode when you start a chat. The AI then sees only column names, types and row counts.
Who it's for
Anyone who works with exports from a CRM, billing system, ad platform or survey tool and needs answers from them: ops, finance, marketing, sales, researchers, founders. You don't need to know SQL or formulas. If you do, the generated SQL is shown so you can check it.
Features
- Files up to around 5 GB and 10 million rows, depending on your laptop's RAM
- CSV and .xlsx support, with each clean sheet opening as its own table
- Questions across multiple files and sheets, so you can join data without VLOOKUP
- Bar, line and distribution charts
- Strict privacy mode
- Free CSV viewer, splitter and CSV-to-Excel converter, no account required
Common problems it solves
1. Excel cut off your rows. Excel stops at 1,048,576 rows and drops the rest without telling you. Sifar Data reads all of them.
2. The file freezes or crashes Excel. Files are read in chunks, so the page stays responsive.
3. The export is a mess. Remove duplicates, fix date formats, split columns and filter out junk rows before they reach a report.
4. The answer is spread across files. Join orders to customers to regions across separate CSVs with one question.
5. The data can't leave your machine. Customer records, payroll and financials stay on your laptop.
Pricing
- Free: 10 messages, all features, no credit card
- 72-hour pass: $6 one-time, 50 messages
- Monthly: $20/month
- Yearly: $129/year (about $10.75/month), 3,000 messages, full refund if cancelled within 14 days
Known limits
Excel sheets need to be standard tables with one header row. Heavy formatting or merged cells can trip it up, and older .xls files need to be saved as .xlsx first. Because no copy of your file is stored, you may need to reconnect it when you come back after a few days.
Feedback
I want to know where it breaks. If you work under strict data rules or with files that crash Excel, try it on a real file and tell me what's missing. You can reach me at hello@sifardata.com, and I'm happy to talk if you'd like to be involved as I build it out.
Explore more artificial intelligence tools on EarlyHunt.
