There's a category of income stream that, for most people without specific technical skills or significant startup capital, used to be effectively off-limits. Not impossible. Not illegal. Just impractical enough that the realistic answer to "can I do this?" was usually "not without things you don't have."
AI has changed the answer for several of them. Not all of them. Some things that seem like they should be easier with AI turn out to require the same human skills they always did. But a few have genuinely shifted — the bottleneck has moved from "you need a skill you don't have" to "you need the judgment to use a tool well."
That's a meaningful difference. Let me walk through the two I think are most significant for the kind of portfolio we're building.
Faceless YouTube.
If you've been on YouTube recently, you've probably noticed a category of channel that produces regular, high-quality video content without any on-camera presenter. Narrated slideshows. Animated explainers. Documentary-style videos built from stock footage, voice-over, and music. Channels in topics like history, true crime, finance, science, and technology that have hundreds of thousands of subscribers and millions of views without a single shot of a human face.
These channels exist because YouTube rewards consistent, well-optimized, engaging content — regardless of whether that content required a camera and a presenter. The income model is YouTube's ad revenue program plus sponsorships, often plus affiliate relationships.
The historical barrier: producing a faceless YouTube video at quality that meets audience expectations required a scriptwriter, a voice-over artist, a video editor, and usually some graphic design support. For a two-person operation building ten income streams, that's not a feasible production pipeline.
The AI unlock: script generation from an outline, AI voice synthesis that sounds genuinely good (not the robotic text-to-speech of five years ago), AI-assisted video editing, and automated thumbnail generation. The pipeline from topic to published video is now manageable by one person with a good system.
I want to be careful here: this doesn't mean it's easy or that the quality threshold has dropped. The channels that perform well in this space still have good research, clear narration, and competent editing. AI makes the production pipeline tractable; it doesn't replace the editorial judgment that makes a channel worth watching. But for someone who has good editorial instincts — the kind of instincts you develop over decades of reading and writing and thinking about what makes a story worth telling — this is now an accessible stream.
Local lead generation.
Local lead gen is a model where you build a website that ranks for a local service query — "emergency plumber in [city]" or "tree removal near [neighborhood]" — and then either sell the leads to service providers or run the site as a directory with paid listings.
This is different from what I described with the directories, though related. The pure lead gen version is more aggressive in its targeting: you're competing directly for high-intent search queries and the income per lead is higher because the service value is higher.
The historical barrier: competitive local SEO required significant expertise, content volume, and time investment. Building a site that ranks for "roofer in [city]" means competing with established local businesses, review platforms, and sometimes franchise networks that have been building domain authority for years. Winning that competition required either significant SEO expertise or significant resources to hire it.
The AI unlock: content production volume at low cost, systematic technical SEO execution, faster research on what's actually ranking and why, and the ability to build and maintain a site with minimal technical overhead. The expertise requirement hasn't disappeared — you still need to understand how search ranking works — but the execution capacity has increased substantially.
What these two have in common.
Both of these streams were always possible in theory. The problem was always execution capacity — you needed a specific combination of skills or resources that most individuals don't have. AI doesn't give you a shortcut past the judgment and strategy. It gives you the production capacity that used to require a team.
That's the pattern I keep coming back to. The judgment has to be yours. The production can be AI-assisted. And for people who have good judgment developed through other disciplines — law, journalism, medicine, teaching, whatever your background is — that's actually a significant advantage in the AI era. You already have the hard part.
This Week in AI: AI voice synthesis has reached a quality threshold in the past year that makes it viable for a wide range of commercial applications — narration, accessibility tools, podcast production. The technology has also raised legitimate questions about consent and verification that the industry is still working through. Both things are true: genuinely useful capability, genuinely complicated implications.
Curious about the full list of streams we've evaluated — including ones we ruled out? Join The Upstream community for the unfiltered version. Details at start.tenstreamslab.com.