9 AI channels · what each is for

What I reach for Edmund's channel for: the Simple · Lovable · Complete scoping discipline, the validation-before-building framework, and his honest framing of slow launches. The thing that makes Edmund's content unusual is that he tells you the apps didn't go viral. They made money slowly. The launch graphs are unsexy. That's the value.

Where the channel is less useful: when he's actively promoting one of his own products in the middle of a build video. The promotion segments aren't dishonest, but they slow the signal-to-noise. Skip those, watch everything else.

If you only watch one Edmund video, watch How I Code Profitable Apps SOLO — it's the framework summary the other videos demonstrate.

The channel covers the specific ergonomics of using Claude Code and Codex in real engineering work — the dashboard, multi-window terminal management, agent configuration. The videos are short, technical, and show real screens. Closest match to what a working engineer would actually want to see about these tools.

Where it goes shallow: less product-thinking, more tool-thinking. If you want what should I build, this isn't the channel. If you want how should I configure the tool I've already decided to use, it's the best on this list.

What I reach for Nate's channel for: rapid coverage of new agentic-coding tooling. When a new feature ships in Claude Code or Codex, Nate usually has a hands-on within a few days. The demos are dense.

Where it gets less useful: the channel skews toward coverage over critical evaluation. Every new tool gets a "this changes everything"-adjacent framing in the title. The titles overpromise; the videos themselves are usually more measured than the thumbnails suggest. I watch the videos with the volume on and the thumbnails ignored.

Useful for: how someone is actually using these tools to ship things, not just demoing the tools in isolation. The videos lean toward case studies of using agents for real projects.

Less useful for: bleeding-edge tool coverage. If a new tool dropped yesterday, Ayush probably won't have a video about it for a week or two. The trade-off is depth over immediacy.

Useful for: breadth. If you want to see what's possible across many agent platforms and integrations, this is the volume channel.

Less useful for: depth. The high volume means a lot of videos are shorter, more demo-y, less analytical. I cherry-pick by topic; I don't watch every upload.

Useful for: the intersection of what to build and how to build it with AI. When the content lands well, it's both. When it doesn't, it's one or the other.

Less useful for: deep technical work. The channel sits closer to the business side of the solo-founder spectrum than the engineering side.

Useful for: the what happens when you let an agent loose on X for an afternoon genre. Light, fast, occasionally surprising.

Less useful for: anyone looking for a repeatable framework. Jack's content is closer to interesting experiments than systematic methodology.

Useful for: the money math behind AI businesses — pricing, infrastructure cost, monetization. The intersection most channels in this list skip.

Less useful for: pure technical content. Jason isn't shipping the apps; he's analyzing them.

Useful for: keeping the broader picture in view. Model releases, company moves, regulatory shifts, the cultural conversation.

Less useful for: anything tactical. This is context, not implementation.

Edmund Yong. Not because his content is the most spectacular — it isn't. Because the discipline it teaches (SLC scoping, validation-before-building, slow honest launches) is the one I'd most regret not internalizing.

The bleeding-edge tooling channels (Chase, Nate) are useful but interchangeable on a 12-month time horizon — there's always a Chase-or-Nate-equivalent covering whichever tool is current. The frameworks Edmund teaches don't go out of date.

The popular AI channels I deliberately stopped watching:

• Channels that pump every new model release with no real test. The thumbnails are GPT-X CHANGES EVERYTHING. They don't. Nothing changes everything in 10 minutes.
• "Agents are about to replace developers" content. They aren't, on the timelines the videos imply. The framing is reliably wrong.
• Pure tutorial channels that cover one tool in a 45-minute walkthrough format. The documentation is faster.
• Anything by an influencer whose audience size doubled in the last six months and who isn't shipping code visibly anywhere.

There's nothing wrong with watching any of those. They just stopped being a productive use of my hours.

The signal I trust most across this list: does the creator ship something themselves? Edmund ships apps. Chase ships configurations. Nate ships demos against his own projects. Ayush ships case studies of his own work.

The signal I trust least: engagement metrics on the channel. A video with 2 million views by someone who hasn't shipped a project in two years is worth less than a video with 8,000 views by someone who shipped this morning. Engagement is gameable; shipping is not.

That's the curation principle the whole list runs on. The watchlist might shift next quarter. The principle won't.