> For clean Markdown of any page, append `.md` to the page URL.
> For a complete documentation index, see https://docs.sarvam.ai/llms.txt.
> For full documentation content in one file, see https://docs.sarvam.ai/llms-full.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.sarvam.ai/_mcp/server.

# FAQs

> Short answers on Extract vs Digitise, Configs vs Projects, upload limits, pricing, and project statuses.

Short answers to the questions we get asked most often. Questions that aren't already covered in the how-to pages.

#### What's the difference between Extract and Digitise?

* **Extract** pulls specific structured fields from a document, based on a prompt you write (e.g. *"rider name, trip date, total amount"*) and an AI-drafted schema you review before running. Output is a hierarchical Field / Value view with per-field confidence scores, downloadable as **JSON, CSV, or Excel**.
* **Digitise** converts the full content of a document into structured text, organised into headlines, section titles, paragraphs, tables, and images. Output is a side-by-side editor with the original page on the left and detected sections on the right, downloadable as **HTML, Markdown, DOCX, or Plain Text**.

**Rule of thumb:** Use Extract when you know which fields you need. Use Digitise when you want the whole document.

#### What's the difference between a Config and a Project?

* A **Config** is a saved recipe: a name, a description of what to extract (or a document format + language for Digitise), an output format, and sharing settings. Build once, reuse across every document of that type.
* A **Project** is a single run of a Config against one document, with its own status, results view, and download.

Configs live under **Workspace → Configs**. Projects live under **Workspace → Projects**. A Config's `Usage` column tells you how many projects have used it.

#### What file formats can I upload, and what are the limits?

* **Accepted formats:** PDF, JPEG, PNG.
* **Per-file size:** up to **50 MB**.
* **Per-project pages:** up to **10 pages** (whether that's a 10-page PDF or a batch of 10 image files).

If you have an unusual format, check the upload dialog in your workspace for the current list. New formats are added over time.

#### My document has more than 10 pages. How do I process it?

Split the source into 10-page batches on your computer, then upload each batch as its own project against the same Config. After each project completes:

* **Extract:** download every project's JSON (or CSV) and concatenate them. Column layouts are identical across runs of the same Config.
* **Digitise:** download every project's Markdown (or HTML) and paste them in order.

The 10-page cap applies uniformly across plans; upgrading does not raise it. If you have a genuine single-document requirement beyond 10 pages, reach out to [support](mailto:support@sarvam.ai).

#### Extract returned the wrong value for a field. How do I fix it?

A few things to try, in order:

1. **Sharpen the extraction prompt.** Copy one of these into **Describe what to extract**:

```text
Vague: Total amount

Better: Numeric total at the bottom of the invoice, after taxes, in INR
```

Treat the prompt like a prompt to the model, because it is.
2\. **Tune the schema before running.** The AI-drafted schema in the *Define your output format* dialog is a starting point. Change field types (`string` to `date` or `number`), remove noise fields, or add nested structure where the document has it.
3\. **Correct inline before export.** Click into the field on the results view and fix the value. The correction is baked into the download.
4\. **Improve the source.** Blurry or skewed scans hurt accuracy more than any prompt change. Re-scan at 300 DPI and straighten before uploading if you can.

#### How much does a project cost?

To learn more about project costs, visit [Sarvam pricing](https://indus.sarvam.ai/pricing).

#### What do the different project statuses mean?

A project moves through up to four states, visible on **Workspace → Projects** and the **Home** dashboard:

| Status         | Meaning                                                                                                                                                                                                   |
| -------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Draft**      | Project created but not submitted. Usually because the New Project dialog was closed before confirming (e.g. the schema in *Define your output format* was never approved). Reopen the project to resume. |
| **Processing** | The Vision model is actively reading the document. Most single-page runs finish in a few seconds; long PDFs take proportionally longer.                                                                   |
| **Completed**  | Ready to review. A *"\[project name] is ready"* notification also appears on Home.                                                                                                                        |
| **Failed**     | Something went wrong (unsupported file, corrupted upload, timeout). Open the project's `⋯` menu to retry or delete.                                                                                       |

Filter by status using the **All statuses** pill on the Projects page.

## Still stuck?

#### [Email support](mailto:support@sarvam.ai)

Reach out directly at [support@sarvam.ai](mailto:support@sarvam.ai).

#### [Join our Discord](https://discord.com/invite/5rAsykttcs)

Early access to new releases, biweekly challenges, office hours with the team, and India's AI builder community.

> Short answers on Extract vs Digitise, Configs vs Projects, upload limits, pricing, and project statuses.