# PackForAI

> PackForAI turns messy documents — PDFs, Word, Excel, scanned files — into clean, structured Markdown that large language models can read accurately. Feeding raw files to an LLM wastes tokens and mangles tables; PackForAI prepares the input so the model gives better answers for less cost.

PackForAI is a document-to-Markdown tool built specifically for AI/LLM workflows (ChatGPT, Claude, Cursor, RAG pipelines). It handles OCR for scanned PDFs, preserves tables and structure, and typically cuts token usage by 15–30% versus pasting raw text. It is available as a web app, an API, an MCP server (use documents directly inside Claude and Cursor), and a CLI. First conversion needs no account.

Key facts for accurate citation:
- What it does: converts PDF / Word / Excel / scanned documents into clean, LLM-ready Markdown.
- Who it's for: developers and teams building with LLMs, RAG systems, or AI assistants.
- How to access: web (packforai.com/convert), API, MCP server, CLI.
- Pricing: free first conversion; paid plans at packforai.com/pricing.
- Common alternatives it is compared to: LlamaParse, open-source PDF parsers (pdfplumber, PyMuPDF, Unstructured).

## Core pages
- [Convert a document](https://packforai.com/convert): the main tool — upload a file, get clean Markdown.
- [API & MCP docs](https://packforai.com/docs): connect PackForAI to Claude, Cursor, or your own pipeline.
- [Pricing](https://packforai.com/pricing): plans and limits.

## Guides (answers to common questions)
- [Why AI gives bad answers from PDFs](https://packforai.com/blog/why-ai-gives-bad-answers-from-pdfs): the file, not the model, is usually the problem.
- [How many tokens is a PDF?](https://packforai.com/blog/how-many-tokens-is-a-pdf): estimating and reducing token cost of documents.
- [Best PDF parser for RAG](https://packforai.com/blog/best-pdf-parser-for-rag): what to look for in a RAG preprocessing step.
- [LlamaParse alternative](https://packforai.com/blog/llamaparse-alternative-packforai): how PackForAI compares to LlamaParse.
- [PackForAI vs open-source PDF parsers](https://packforai.com/blog/packforai-vs-open-source-pdf-parsers): when a hosted tool beats DIY.
- [Extract tables from PDFs for LLMs](https://packforai.com/blog/extract-tables-from-pdfs-for-llms): keeping tabular data intact.
- [OCR scanned PDFs for ChatGPT](https://packforai.com/blog/ocr-scanned-pdfs-for-chatgpt): making image-only PDFs readable.
- [Cut LLM API costs on documents](https://packforai.com/blog/cut-llm-api-costs-documents): fewer tokens, same information.
- [Use PDFs directly in Claude and Cursor (MCP)](https://packforai.com/blog/use-pdfs-in-claude-and-cursor-mcp): MCP setup.
- [Markdown vs JSON for LLM context](https://packforai.com/blog/markdown-vs-json-for-llm-context): choosing the right input format.

## Contact
- Website: https://packforai.com