Own HackerTube 上线 AI 个性化视频策展工具

原标题:Own Your HackerTube,an AI Curator Inspired by Karpathy's HackerTube
45AI 研判分

Own HackerTube 推出了一款受 Karpathy 启发的 AI 视频策展工具,允许用户通过自然语言定义偏好标准。该工具能自动从关注的 YouTube 频道中筛选高质量内容,并基于用户反馈持续学习优化推荐结果,最终生成独立的个人化视频站点。

报道正文

本文目录32 个章节

Quality input, filtered by AI and tuned to your taste. Follow YouTube channels, write your standards, and an AI picks only the videos worth your time, then learns from what you like and don't, all on your own YouTube-style site.

Create your site — freeSee a live example

See it live

This is a real site. Its owner follows nine YouTube channels and told the AI what they care about. Every video below was picked by the AI.

Their standards, in their own words

I'm a small-team AI founder. I like watching interviews with overseas AI founders, especially about high-level thinking, hands-on practice and AI-native startup experience. What I don't want: clickbait titles, and videos that only explain concepts or give opinions without real cases, numbers or concrete steps.

[

2:34

和曾鸣聊产业史观:残酷的真相、会消亡的公司、优秀≠卓越、“OAI、Anth大概率不是原生时代大赢家”

张小珺商业访谈录

165.7K views • 2 months ago

高质量UP主](https://ownhackertube.cc/s/ai-picks/i/31d2b7f8-ca6a-46a1-a8ef-cc8f972f9a18)[![](https://i1.hdslb.com/bfs/archive/ffa60fa476e638a522128ee9d337b53e4c90b372.jpg)

2:51

对Dyna创始人Lindon Gao的3小时访谈:机器人世界的迷雾、分歧与抉择,修心、《沉思录》与王阳明

张小珺商业访谈录

63.4K views • 2 weeks ago

喜欢的高质量UP主](https://ownhackertube.cc/s/ai-picks/i/cde9b45e-9a3b-44d4-8e35-737d16bc915d)[![](https://i2.hdslb.com/bfs/archive/0dc5da5b59cd92ef1d6021aa3d166f7c0dd4fe66.jpg)

1:59

和科沃斯钱程聊,父与子、企二代接班的样本与肌理、机器人、我内心的追寻

张小珺商业访谈录

60.1K views • 1 week ago

这个UP主的采访质量还是蛮高的](https://ownhackertube.cc/s/ai-picks/i/fdd921f3-4f5f-47b7-9d64-ed405c59465e)[![](/api/v1/file/hot_img_78c0a70c68cb6b1d24e613c14c3df4ee.webp)

38

Taking on ServiceNow: Coding Agents for the Rest of Us with Serval Co-Founder and CEO Jake Stauch

No Priors: AI, Machine Learning, Tech, & Startups

3.2K views • 4 days ago

AI pickStrong fit for a small-team AI founder: an overseas AI startup founder explains how they’re productizing coding agents for enterprise IT, with concrete guardrails, workflow design, and go-to-market lessons from attacking a big incumbent. It also covers how AI changes the product and operating model.](https://ownhackertube.cc/s/ai-picks/i/1721fb97-2058-42ca-90c3-2978c09eb80a)[![](/api/v1/file/hot_img_abacdccbfe94b45a232a76121bf3e0ad.webp)

39

The AI Billing Stack for Healthcare

YC Root Access

4.1K views • 6 days ago

AI pickFounder interview with concrete startup execution: finding a wedge, winning the first customer, and scaling an AI-assisted workflow business in a huge legacy market. Useful if you want practical lessons on PMF, automation, and where agents actually fit.](https://ownhackertube.cc/s/ai-picks/i/52a48eb6-9110-4960-8e09-1fefd1707052)[![](/api/v1/file/hot_img_1f5e615b2399597a8248dd154c70f8a7.webp)

35

I tested 4 AI agents on 9 real tasks. Here’s who won.

How I AI

239.9K views • 5 days ago

AI pickPractical benchmark of current AI agents on real founder-like workflows, with concrete takeaways on what works for work vs. life. Useful for an AI founder interested in hands-on agent capability rather than theory.](https://ownhackertube.cc/s/ai-picks/i/399a4c46-e592-4a90-84a4-368c9042e5cd)[![](/api/v1/file/hot_img_311522dead6135ed45a2501d55810556.webp)

31

Building AI for the $650B Home Services Market

YC Root Access

18K views • 4 days ago

AI pickStrong founder interview with concrete lessons on market selection, customer discovery, and finding an AI wedge in a real business. The best parts are the hands-on path from failed startup to PMF and scaling a vertical AI product.](https://ownhackertube.cc/s/ai-picks/i/132a1cbe-b69e-4b3a-bef7-3b1971d8d169)[![](/api/v1/file/hot_img_8d6d0825dca6f5a8db22b2c3c0be1f59.webp)

16

What 90,000 Rollouts Reveal About AI Agents Doing Science - Ian Quigley, Leash Bio

Latent Space

13.2K views • 1 week ago

AI pickFounder interview with an AI-native builder on how scientific agents are evaluated in real environments, with concrete lessons on model behavior, mistakes, and what changed in a year. Useful if you want hands-on thinking from an overseas AI startup context.](https://ownhackertube.cc/s/ai-picks/i/a179765f-a4f1-400a-bd6e-3c6777e60da3)[Open the full site →](https://ownhackertube.cc/s/ai-picks)

How it works

1

Say what you care about

Write your standards in plain words and follow the YouTube channels you trust. Videos you add yourself, with a reason, teach it your taste.

The AI picks panel: write your standards in plain words and add the YouTube channels you follow

2

The AI watches, so you don't have to

Every day it checks new videos from your channels and scores them against your standards. Each pick comes with why it is worth your time and the key moments to jump to.

A video page: the reason the AI picked it, and key moments taken from the creator's chapters, each with a short note

3

Your feedback trains it

Picks go live as soon as they pass the filter. You can still tell it which videos you liked and which you did not, and why. The AI remembers your judgment and adds it to its criteria, so over time you get quality plus your own taste.

Your own YouTube-style site: a grid of videos, each with the AI's reason under it

Not another AI summary

Summaries try to replace watching. Filtering decides what is worth watching in the first place.

AI summary toolsOwn Your HackerTubeReplace watching the videoDecide what is worth your time, then point to the key momentsYou paste one video at a timeFollows your channels and checks new videos every dayPrivate notes only you seePublished as your own YouTube-style siteThe same summary for everyoneJudged against your standards, with a reason for every pick

Better input, not a shorter summary.

Most AI video tools summarize a video so you never watch it. This one filters instead: it checks every new video against your standards, so what reaches you is worth your time. You still watch it yourself, and remember it.

  • • Key moments come from the creator's own chapters, never invented
  • • AI picks are labelled as AI; your own words stay yours
  • • Tell it what you like and don't, and why; the AI remembers, and you can change your standards any time

Everything you need for better input

A YouTube-style site

Home grid, watch page, search, and Liked, History and Watch later kept in the reader's own browser.

Key moments

Jump straight to the part worth watching, using the creator's own chapters.

Checks every morning

No prompts to write and no tasks to schedule. New videos from your channels are judged once a day.

It learns your taste

Mark a pick as great, or as not a fit and say why. The AI remembers your judgment and uses it in every future check.

Your own address

Your site lives at ownhackertube.cc/s/yourname, is readable by search engines, and has an RSS feed.

Free to start

One channel and 30 AI checks a month on the free plan. Pro follows up to 20 channels.

Who it's for

Anyone who learns on purpose and guards their attention

In the AI era, good output starts with good input. And a lot of what gets written with AI these days is, frankly, filler. If you care what goes into your head, let a filter that knows your standards do the first pass.

Founders and builders

Keep up with the interviews and talks that matter in your field without scrubbing through hours of footage.

Self-learners

Follow the lecture and course channels you trust and get a short list of what to watch next.

Communities and newsletters

Give your audience a living, always-fresh "watch this" page with your name on it.

Simple pricing

Start free. Upgrade when you want to follow more channels.

Free

$0

1 channel · 30 AI checks a month · up to 50 videos

Pro

$8/month

20 channels · 600 AI checks a month · up to 5,000 videos

or $59 a year

See full pricing

Questions

Is it free?+

Yes. The free plan gives you a site, one followed channel, 30 AI checks a month and up to 50 videos. Pro follows up to 20 channels with 600 checks a month.

Does the AI summarize the videos?+

No. It decides whether a video is worth your time and tells you why in a sentence or two, then points you to the key moments. You still watch the video.

Where do the key moments come from?+

From the chapters the creator wrote in the video description. The AI only chooses among them, so a timestamp is never invented. Videos without chapters still get a pick and a reason, just no key moments.

Which videos and channels work?+

YouTube videos and channels, followed automatically every day. Bilibili is supported too (beta, still being tested): add a video link, or sync an uploader with one click from your own browser, because Bilibili only shows video lists to real browsers. Videos play in each platform's own player and stay there.

What if the AI picks something I don't want?+

Mark it as not a fit and pick a reason, such as too basic, off topic or clickbait. It leaves your site, and the AI remembers your judgment for future picks. You can also rewrite your standards at any time.

Do my readers need an account?+

No. Anyone can open your site. Their Liked, History and Watch later lists stay in their own browser.

How does it decide what is worth watching?+

You describe what you care about in plain words. The AI scores each new video from 0 to 10 against that and publishes the ones that score 7 or more.

Can I use my own domain?+

Custom domains are rolling out for Pro. Until then your site lives at ownhackertube.cc/s/yourname.

Quality input, on autopilot

Describe what you care about, follow a few channels, and let the AI filter. Correct it, and it learns your taste.