Image, PDF & Video

This is the cluster where you upload things, and it's the one where privacy matters most — so here's the promise up front: your files are processed in memory, deleted within 15 minutes of upload, never backed up, never shared, and never used to train anything. There is no file library here. Once you've downloaded your result (or the 15 minutes have passed), the file is gone.

What's here now:

• Image to Prompt — upload any picture and get two ready-to-paste prompts (concise and detailed) for Midjourney, Stable Diffusion, Flux, or any chat-based image tool.
• PDF to Markdown — turn papers, manuals and reports into clean Markdown with headings, lists, tables and LaTeX formulas intact, ready for Obsidian, Notion or GitHub.

More are on the way — PDF summaries, quizzes from documents, handwriting recognition, video summaries — each one announced when it clears testing, not before.

The theme is small, annoying, occasional tasks — the kind that used to mean downloading sketchy shareware or handing a sensitive document to a random converter site with a privacy policy written in 2011. Here an AI model reads it, returns structured output, and forgets it. Every tool runs on the same deal: no account, free, and a deletion clock that starts the moment your upload lands.

Why these two first? Because they're the jobs people search for at midnight before a deadline: "what prompt makes this image" and "get this PDF into my notes without retyping it."

A practical tip: process your files and download the results in the same sitting. The 15-minute window is generous for a single job, but it's a real deadline, not marketing copy. And if the output is headed somewhere permanent — a docs site, your thesis notes — spend thirty seconds spot-checking headings and tables against the original. Conversion is a machine doing its best; verification is you doing yours.