LovedDocs
Open a tool

Prep for AI

Get your text ready for ChatGPT and Claude

Pasting raw documents into an LLM quietly costs you, on tokens, on context limits, and on privacy. These free, browser-only tools fix all three. Work through them in order, or jump to the one you need. Nothing you paste is uploaded.

  1. 1

    Get clean text out of your files

    Strip the formatting noise from documents so the model reads prose, not markup.

    DOCX / PPTX / XLSX to Text

    Word, PowerPoint & Excel → clean, labelled text.

    PDF to AI-ready text

    De-hyphenate and de-clutter PDFs for LLMs.

  2. 2

    Read text trapped in images

    Screenshots, scans and photos are opaque to an LLM until you OCR them.

    Image to Text (OCR)

    Pull text from any image, on-device.

  3. 3

    Pull text from a web page

    Turn an article into clean Markdown without the nav, ads and scripts.

    URL to Markdown

    Any public web page → tidy Markdown.

  4. 4

    Strip hidden junk from pasted text

    Copying from PDFs and Word leaves zero-width spaces and control characters that quietly cost tokens.

    Clean Text

    Remove invisible and look-alike characters.

  5. 5

    Count tokens and cost before you send

    Tokens, not words, decide the price and whether your text fits the context window.

    LLM Token Counter

    Exact GPT tokens + per-model cost and fit.

  6. 6

    Trim or split what doesn't fit

    Cut to the gist, or keep everything and feed it in pieces for retrieval.

    Text Summarizer

    Condense to the key sentences.

    Text Chunker for RAG

    Token-bounded chunks with overlap, for RAG.

  7. 7

    Scrub anything sensitive

    Redact PII and secrets before the text ever leaves your machine.

    Redact PII & Secrets

    Mask emails, cards, keys and more.

  8. 8

    Assemble and reuse your prompt

    Drop the cleaned text into a reusable template with variables, and see the final token count.

    Prompt Template Manager

    Reusable {{variable}} prompts, saved locally.

Go deeper

Questions

Does any of my text get uploaded?
No. Every tool processes your text on your own device. The one exception is URL to Markdown, which fetches the public web page you name through a small server hop, your own text and files are never uploaded.
Is it really free?
Yes, all of these tools are free and need no account. They run in your browser using on-device processing, so there's no API cost to pass on to you.
Why prep text at all instead of pasting it raw?
Three reasons: you pay for tokens you didn't need to send, you can blow past the model's context window and silently lose content, and raw logs or exports often contain personal data you didn't mean to share.