OpenTube

How I Watch ANY Video with AI in Seconds

Brad Bonanno | AI Automation

53,865 views

JOIN MY BOOTCAMP: https://bootcamp.bradbonanno.com
SKILL + SETUP GUIDE: https://brad-b.kit.com/5d6e282f80

AI can now watch videos for free without burning your usage limit. Watch is a free agent skill that gives Claude or ChatGPT a way to understand video, and this version plugs straight into Google's agentic video model, which got more answers right in Google's own test while using 87% fewer tokens. You hand it a link, Gemini does the watching on Google's free API tier, and your AI gets the findings back with timestamps instead of hundreds of screenshots.

In this video I show the three best ways I use it: finding one tiny detail buried in a three and a half hour video, lining up my retention drops with what I was actually saying and showing at that moment, and pulling notes from videos straight into my LLM wiki. Then I walk through the full install in Claude and ChatGPT, getting your free Gemini key, updating if you already have Watch, and when I still use the local engine instead.

Timestamps
0:30 - How it works
1:38 - Use case 1
2:31 - Use case 2
3:53 - Use case 3
4:30 - Setup
5:59 - Updating
6:23 - Local mode
7:07 - Which engine to use

20 Comments

@santvar380 4 days ago

I took a similar/manual approach in Gemini. I created a Gem to extract the video transcript, remove contextual noise (greetings, ads, etc.), and generate a downloadable Markdown file with the data. From there, I either send it to Claude or keep it in Gemini for further analysis, slide creation, etc

@kazanouva007 8 hours ago

Thanks Brad, it works just fine.

@cyberintell2958 4 days ago

Awesome. Keep those informative videos coming. I spend a lot of time watching YouTube videos to learn nearly everyday

@ADDIE. 4 days ago

U always have really good videos men. Awesome thank you

@Aware-Architect 4 days ago

Thank you Brad!

@nekk06 2 days ago

Thank you for thIs!

@DanKnowlson 2 days ago

A great update 👍

@cliftontv 3 days ago

Thanks, Brad. I had your previous versions. I created multiple watch skills geared toward different goals. Some for business. Others for specific types of summaries. For this, I will add the two skills. One that has Claude as the backend and the other with Gemini so I won't have to switch back and forth.

Appreciate you.

@Yuzukhane 4 days ago

watched this video through /watch thank you :D

@joshuakrilov4042 4 days ago

One of my most used and useful skills🤖

@markksantos 4 days ago

i love this channel

@netgaga 4 days ago

excellent video

@VeddedPtyLtd 3 days ago

You are the GOAT

@CarlosSmith-ng7yk 4 days ago

Very valuable skill. Installing after posting this comment.

@AzraelPhalanx 4 days ago

Thank you, Brad. Your tutorial on Claude 'watch video' skill was very helpful. I appreciate it. ⚔

@franky12 1 day ago

You've been able to do that with Google's NotebookLM for years now.

@user-te6ru8kw3h 3 days ago

It takes me about thirty seconds to open the transcript of a video, copy pasta into ChatGpt, and ask it to give me a "breakdown and hot take". SInce it knows what we're always talking about, it puts the video in contexts that I'm interested in. Not as agentic as possible, but it works for me.

@Joey-f2q 2 days ago

Gemini seems good enough

@ProductiveWatcher 3 days ago

Your brilliant! Seen a few of your vids.. The problem with this level of ai use for me is that I dont have (A thriving buisness right now).. or many things to seemly apply ai to in my life and i find once i try something out more in depth, I over analyse, go to deep and it becomes an energy drag where i could be using super simple chats for most things ! I did start building an obsidian 'second brain' But it was a distraction from the simple natutre of the problems that i had to overcome in my life. An Ai second brain would be super cool however, if it functions smoothly you can dump what you learn ect events, meetings discuions and absorb them to the brain!! Ive not ironed it out one day ill probably go back to it!