Show HN: Video2docs – Turn Screen Recordings into Step-by-Step Instructions

[angelina200]

We recently automated a task that used to take some manual hours and was pretty boring itself - documentation and user how-to instructions creation, since before the ability to use AI technology, this was pretty time-consuming thing to do. Now it takes around 15 minutes.

The workflow is simple: record your screen of a task, workflow or feature walkthrough (even without audio) -> get a step-by-step guide with titles, descriptions, and contextual screenshots out of it automatically, auto-translate, and edit the output if needed as in a regular doc before final export. How it works? by analyzing unique video frames via LLM and then combining that information into final docs.

This automates the process and makes documentation much less painful, because you can start with a solid base and refine if needed instead of a blank page. We built our own tool based on this workflow, and maybe will be useful for others, too.

It is free to try, and we are happy to hear any feedback! Thank you.

[legitimate_key]

This is a smart workflow. I've been doing something similar (record screen, then manually write docs) and the AI approach saves hours.

One thing that still trips me up though - the prep before hitting record. I spend like 10 minutes closing personal tabs, clearing browser history, making sure nothing sensitive is visible on screen. By the time I'm "ready" to record, I've lost the spontaneous energy that makes demos feel natural.

Does your tool handle any of that? Like auto-detecting sensitive content before processing or helping sanitize the recording after? Or is it assumed you're recording in a clean environment?

Curious because the actual documentation generation is only half the workflow. The "setup tax" before recording is the part that kills my momentum.

Congrats on shipping this. The 15 minute turnaround is impressive.

[angelina200]

Thank you! yes, totally agree, but i think it is more important in case of demo videos and not the process recording -> documentation, where the written output is the biggest value.

That said, i still agree there might be some sensitive parts that gets into screenshots of the documentation, so in the future we plan to add some sensitivity detection and screenshot blurring options, but for now yes it is better to prepare clean environment. Thanks again for the thoughtful comment!