Video to blog post

A tutorial is only useful if every step is right, so the tool never presents a guess as something your video said. For each step, the AI must point to the transcript line where you said it and copy your words. The tool then looks for those words in your transcript. If they are there, the step gets that moment’s time. If they are not, the step is marked “Not found in the transcript” and has no time.

Some actions happen only on screen: you click a button while saying “now this one here”. The AI is told not to guess the label. It writes the step in general words and marks it “Only shown on screen”, so you can watch that moment and add the real name. A notice at the top counts the steps left to check, and the marks stay in the Markdown until you tick them, so a guess is still labelled if you share the draft.

Once your recording is added, each step offers a screenshot at the time it was said. Press it and the tool takes that frame from the video on your device. If the moment is a little early or late, play or scrub the video to the right frame, pause, and press Use paused frame. You choose every image. Nothing is captured until you ask, and you can remove any screenshot.

Screenshots are saved as JPEG, up to 1920 pixels wide, and named by step number, such as step-03.jpg, so a title with odd characters can never produce a broken file name. If you started from a transcript, add the recording after the tutorial is written. Screenshots live only in this browser tab, so download before you close it.

Download .zip gives you tutorial.md and an images folder. The Markdown uses plain headings, numbered steps and standard image links, so it works in WordPress, Ghost, Notion, GitHub, Docusaurus and most static site generators. Copy Markdown copies the text alone. Each step ends with its time in the video. Paste your YouTube link under Link timestamps to YouTube and each time becomes a link that opens the video at that moment.

Before you publish, read the article once with the software open. Check names of buttons and menus, and add anything you showed but did not say. A written guide is also a good place to add the exact values, links and code that are hard to read in a video.

You can write the guide by hand while replaying the video, pausing at each step to type it out and take a screenshot. It is slower, but you see every frame; the extract frames tool saves any frame as a full-size image for those screenshots. If you only need the words, the video to text converter gives you a full transcript to edit. A general AI chat assistant can also turn a transcript into an article, but it will not tie steps to times, mark guesses or take screenshots from your video.

For a YouTube upload, the same transcript can make your chapters and description. For planning the next recording, read how to make a tutorial video. To choose between a recorded walkthrough and a click-by-click guide from Scribe or Tango, see Rescreen vs Scribe.

The tool works on recordings up to two hours long, with up to 120,000 characters of transcript. Long videos are split into parts that are written at the same time and joined, so steps near the end are kept. It is free with a limit of 10 tutorials an hour from one connection, and a daily limit across all visitors, so it can be busy at times. If it cannot finish, your transcript stays on the page and you can try again.

It works in current versions of Chrome, Edge, Safari and Firefox on a computer. Transcribing a recording on your device works best in Chrome or Edge; in other browsers, paste a transcript or open a subtitle file. Screenshots need a video your browser can play. It cannot read a YouTube link or publish to your blog.