What happens to nested objects inside a JSON record?

A nested object or array within a field is kept as-is in that cell rather than auto-split into separate columns, so deeply nested JSON is easiest to work with if you flatten it before uploading.

What if my file isn't an array at the top level?

A single JSON object (not wrapped in an array) is converted into a one-row table, one column per top-level key.

Does this use AI, like the PDF and image tools?

No. JSON is already a structured format, so it's parsed deterministically, the same way CSV, XLSX, and XLS are.

What if my JSON is an object that contains an array?

If the top level is an object with an array inside it (for example { "orders": [ ... ] }), that array becomes the table rows. Any other top-level keys are kept alongside as key-value details instead of being dropped.

Which keys become CSV columns?

Columns are taken from the keys of the first object in the array. A key that only appears in later objects is not added as a column, so put the most complete record first or normalise the file before uploading.

What happens to a JSON array of plain values, like [1, 2, 3]?

Each value becomes its own row in a single column named value, because there are no object keys to use as column names.

How accurate is the AI data extraction?

Datafromfile uses AI vision models to read document structure (tables, totals, and line items) directly from PDFs, scans, and photos. Extraction quality varies by document: a clean digital PDF extracts more reliably than a blurry photo, and clearly printed text is more reliable than handwriting or a heavily irregular table layout. Every result is flagged as either high confidence or needs review, never a fabricated precision percentage, and you can check the data yourself in the editable workspace before exporting. Every plan uses the same extraction models; paid plans get more monthly volume and faster concurrent processing, not a different or more accurate model.

What file types can I upload, and what can I export to?

Upload PDF, JPG, PNG, WebP, XLSX, XLS, CSV, JSON, XML, or HTML. Every plan, including Free, exports to Excel (.xlsx), CSV and JSON; Pro and above add Markdown export too, see the plan comparison on the Pricing page for the exact breakdown.

What if my file is too large, or has a lot of pages?

Each plan has its own real per-file limit, enforced on the actual file you upload, not just a claim: 3 MB on Free, 10 MB on Pro, 25 MB on Business, and 100 MB on Enterprise. A multi-page document, like a 10-page bank statement, is supported and read as one continuous result as long as the whole file stays under your plan's limit; a scan larger than that may need to be compressed or split first.

Is my financial and business data private and secure?

Yes. Each user’s documents are isolated at the database level, every upload travels over TLS, the original file is not kept once extraction finishes, and nothing you upload is used to train AI models.

Is the Free plan really free, and what actually triggers an upgrade?

Yes: 15 real AI operations a month (an extraction, an Ask Your Data question, or an AI transform each count as one), no credit card required, indefinitely. Excel export and AI-powered cleaning are included on Free. You'd look at a paid plan once you need more than 15 operations a month, bigger files, larger batch uploads, or Markdown export, all Pro and above.

Convert JSON to CSV Online

Convert a JSON array or object into a clean CSV file: deterministic parsing, no AI model involved, ready for Excel, Google Sheets, or a pandas import.

Drop your file here

or browse your computer

JSON · up to 3 MB

Features

Deterministic parsing, not guessed by an AI model, so the same input always produces the same output
A JSON array of objects becomes one CSV row per item, columns from the object keys
A single JSON object becomes a one-row property/value table
Review and edit every cell in the workspace before exporting

How it works

Upload your JSON: A .json file containing an array of records or a single object.
Automatic structure parsing: Array items become rows; object keys become columns.
Export CSV: Download as .csv or copy straight to your clipboard.

Frequently asked questions

What happens to nested objects inside a JSON record? A nested object or array within a field is kept as-is in that cell rather than auto-split into separate columns, so deeply nested JSON is easiest to work with if you flatten it before uploading.
What if my file isn't an array at the top level? A single JSON object (not wrapped in an array) is converted into a one-row table, one column per top-level key.
Does this use AI, like the PDF and image tools? No. JSON is already a structured format, so it's parsed deterministically, the same way CSV, XLSX, and XLS are.
What if my JSON is an object that contains an array? If the top level is an object with an array inside it (for example { "orders": [ ... ] }), that array becomes the table rows. Any other top-level keys are kept alongside as key-value details instead of being dropped.
Which keys become CSV columns? Columns are taken from the keys of the first object in the array. A key that only appears in later objects is not added as a column, so put the most complete record first or normalise the file before uploading.
What happens to a JSON array of plain values, like [1, 2, 3]? Each value becomes its own row in a single column named value, because there are no object keys to use as column names.
How accurate is the AI data extraction? Datafromfile uses AI vision models to read document structure (tables, totals, and line items) directly from PDFs, scans, and photos. Extraction quality varies by document: a clean digital PDF extracts more reliably than a blurry photo, and clearly printed text is more reliable than handwriting or a heavily irregular table layout. Every result is flagged as either high confidence or needs review, never a fabricated precision percentage, and you can check the data yourself in the editable workspace before exporting. Every plan uses the same extraction models; paid plans get more monthly volume and faster concurrent processing, not a different or more accurate model.
What file types can I upload, and what can I export to? Upload PDF, JPG, PNG, WebP, XLSX, XLS, CSV, JSON, XML, or HTML. Every plan, including Free, exports to Excel (.xlsx), CSV and JSON; Pro and above add Markdown export too, see the plan comparison on the Pricing page for the exact breakdown.
What if my file is too large, or has a lot of pages? Each plan has its own real per-file limit, enforced on the actual file you upload, not just a claim: 3 MB on Free, 10 MB on Pro, 25 MB on Business, and 100 MB on Enterprise. A multi-page document, like a 10-page bank statement, is supported and read as one continuous result as long as the whole file stays under your plan's limit; a scan larger than that may need to be compressed or split first.
Is my financial and business data private and secure? Yes. Each user’s documents are isolated at the database level, every upload travels over TLS, the original file is not kept once extraction finishes, and nothing you upload is used to train AI models.
Is the Free plan really free, and what actually triggers an upgrade? Yes: 15 real AI operations a month (an extraction, an Ask Your Data question, or an AI transform each count as one), no credit card required, indefinitely. Excel export and AI-powered cleaning are included on Free. You'd look at a paid plan once you need more than 15 operations a month, bigger files, larger batch uploads, or Markdown export, all Pro and above.

Related converters

Excel, CSV & JSON Data Extractor — Clean and convert Excel, CSV and JSON data online. Fix messy headers and merged cells, then export to XLSX, CSV or JSON: deterministic parsing, no AI needed.
PDF to Excel Converter — Convert PDF to Excel online with AI. Extract tables from native and scanned PDFs into an Excel-ready spreadsheet with intact columns and numbers.
Image / Screenshot to Excel — Convert a photo or screenshot of a table into an editable Excel spreadsheet. AI reads JPG, PNG and WebP images, including skewed and handwritten tables.
PDF to CSV Converter — Convert PDF to CSV online. Extract clean, structured tabular data from PDF documents into comma-separated values ready for Excel and pandas.
Image to CSV Converter — Convert an image to CSV with AI table recognition. Turn photos and screenshots of tables into clean comma-separated values, even with uneven lighting.
Image OCR: Extract Text from Photos — Extract text and data from images with AI-powered OCR. Upload a JPG, PNG, or WebP photo or screenshot and get structured, editable data. Free to try.