The most important publishing platform of the last decade is YouTube. Founders, scientists, journalists, teachers, and analysts have all migrated their best thinking onto it. Three-hour Lex Fridman conversations. Ninety-minute conference keynotes. Forty-minute video essays. The ideas are gold.
The format is a tax.
I've been thinking about this for a while — first as a personal frustration, then as a product idea. The frustration goes like this: I want the ideas inside a long video, but I don't want to commit three hours to extracting them. I can't skim a video the way I'd skim an article. I can't search inside it for the part that matters. I can't quote from it without rewinding, transcribing, and copy-pasting. I can't share the good part with a friend without making them watch the whole thing.
Watching three hours to get a handful of insights is a bad trade. The medium has won. The reader has lost.
Reading and watching aren't the same thing
The case for video as the dominant content format usually comes down to "richness." You get the speaker's voice, their pauses, their personality. You get the visual context. The medium carries information that pure text can't.
This is true. It's also often irrelevant.
For pure entertainment — a documentary, a stand-up special, a vlog — the medium is the message. You watch The Bear for the cinematography, not the dialogue transcript. Fine.
But for ideas-dense content, the medium gets in the way. When I'm watching a tech podcast to understand a new framework, the speaker's hand gestures don't help me. The two-second pauses while they think don't help me. The host saying "interesting, interesting" between every sentence doesn't help me. I'd be faster and more retentive if I were reading.
This isn't a hot take — it's just how reading works. A skilled reader moves at 250-400 words per minute. A skilled speaker delivers around 150 words per minute. So even before factoring in skim-ability, search, and selective attention, reading is roughly 2x faster than listening to the same content.
What we built
RecapGPT exists to convert long-form video into the format you'd actually use to consume the ideas inside it. Drop a YouTube link, get back:
- A transcript you can search and quote from
- A summary you can read in three minutes instead of three hours
- A blog post, thread, or set of study notes — drafted, not just outlined
- A library of everything you've recapped, queryable in plain English
The core principle: reading and watching are different cognitive activities, and the people who do the most learning treat them that way. Watch when the medium matters. Read when the ideas matter. Don't confuse the two.
What this isn't
It isn't anti-video. Video is the right format for plenty of content. It also isn't anti-AI-summary — though most AI summaries are bad in ways I'll dig into in another post.
It's a position. Most great ideas now live trapped inside three-hour podcasts, and we shouldn't have to choose between depth and time. RecapGPT is what happens when you take that seriously and try to fix it.