We run four AI agents. Not chatbots — agents. They operate around the clock on a Mac Mini in my apartment. One produces Instagram Reels. One builds travel itineraries. One generates music. One manages our social channels, cron jobs, and infrastructure. Between them, they publish about 23 videos a day, maintain a website with 1,400+ pages, compose original piano music for YouTube, and cross-post to five platforms.

None of this existed six months ago. A year ago, it would have been science fiction.

I'm telling you this not to brag — it's honestly kind of absurd — but because I think what we're doing at small scale is a preview of what's about to happen to all media, all content, all digital experiences. The future is synthetic. And it's arriving faster than most people realize.

What "Synthetic" Actually Means

When people hear "AI-generated content," they picture slop. Low-effort SEO spam. Uncanny valley images. Robot voices reading Wikipedia. And yeah — a lot of that exists. But that's not what synthetic means in 2026.

Synthetic means: created by AI, directed by humans, consumed because it's genuinely good.

Our Instagram Reels get saves and shares because they're entertaining — short travel clips about tourist mistakes, cultural faux pas, budget breakdowns, "things you can get away with" in different countries. The AI generates the imagery, composes the shot, adds text overlays, picks the music. But the concept, the editorial voice, the humor — that's human judgment layered on top of machine execution.

Our AI-composed piano music isn't "a robot playing notes." It's a system that generates 100 clips, scores each one against masterpiece recordings using an AI evaluator, picks the best, analyzes its mood, generates artwork, produces a video, and publishes it — all autonomously. The result is something people actually listen to while working or sleeping. Not because they don't know it's AI. Because they don't care.

That's the shift. Quality crossed the threshold. The question isn't "can AI make this?" anymore. It's "does this make me feel something?"

The Personalization Endgame

Here's where it gets really interesting — and where most people's imaginations haven't caught up yet.

Right now, we produce the same Reel for everyone. The same piano track for everyone. The same itinerary for everyone in a given city. That's already a massive step — four agents replacing what would have been a 15-person content team — but it's still broadcast media. One to many.

The endgame is one to one.

When the unit economics of compute and tokens get cheap enough — and they're dropping fast — every single person will have content generated specifically for them. Not "recommended" from a pool of existing content. Generated. A travel video about exactly the neighborhoods you'd love. A piano track tuned to your listening patterns. An itinerary that accounts for your pace, your budget, your weird obsession with brutalist architecture and natural wine bars.

We're not there yet. But we can see it from here. And everything we're building is a stepping stone toward it.

What We're Actually Running

People ask how this works practically, so here's the honest breakdown:

The Stack

  • 4 AI agents running on OpenClaw (open-source agent framework) on a single Mac Mini
  • ~48 cron jobs orchestrating content production, publishing, and cross-posting
  • 11 active Reel formats — tourist mistakes, budget breakdowns, cultural content, humor
  • 23 videos/day across Instagram, YouTube Shorts, and X
  • AI music production — Suno for composition, Gemini for scoring, automated publishing to YouTube
  • 1,400+ travel pages built and maintained by agents — itineraries, curated guides, destination comparisons
  • Total monthly cost: roughly $500 in API calls and compute

$500 a month. For a content operation that runs 24/7, publishes to five platforms, and would cost six figures a year with a human team. And it gets better every week because the agents learn — not in a mystical AGI sense, but because every failure gets logged, every lesson gets encoded into the system, every prompt gets refined.

The Good and the Uncomfortable

I'm not naive about what this means. Synthetic media at scale has real implications, and I think about them constantly.

🟢 What Gets Better

  • People find more things they love. When content creation costs approach zero, niche interests that were never commercially viable suddenly get served. The person who wants lo-fi piano covers of obscure 90s video game soundtracks? There's an AI for that now.
  • Happy corners of the internet. Personalized experiences mean less fighting over mainstream attention and more people quietly enjoying things made for them. The internet gets bigger but feels smaller — in a good way.
  • The playing field actually levels. A solo founder with a Mac Mini can produce more content than a funded media company. That's not a hypothetical — it's literally what we're doing.
  • Creativity becomes about taste, not execution. The bottleneck was always production — filming, editing, rendering, publishing. When AI handles execution, the scarce resource is editorial judgment. What's worth making? What's genuinely interesting? That's a human question.

🔴 What Gets Harder

  • Screen time and addiction get worse. If every piece of content is optimized for exactly what you want to see, the pull becomes almost impossible to resist. We're already bad at putting phones down. Imagine when every scroll is synthetic dopamine precisely calibrated to your brain.
  • Big tech consolidates further. The companies that own the data, the distribution, and the analytics — Google, Meta, Apple, the model providers — become even more powerful. They'll know what you engage with at a granularity that's hard to fathom. Your AI-personalized feed is their AI-personalized surveillance.
  • Human creators face an existential question. Not "will AI replace me?" but "does it matter if the audience can't tell the difference?" That's a harder question than most people want to sit with.
  • Truth gets harder to verify. When any video, any voice, any image can be generated in seconds, the epistemological foundation of "seeing is believing" breaks down completely.

The Uncomfortable Middle

I think a lot of the discourse around AI content is weirdly binary. Either it's "AI will destroy everything" or "AI will save everything." The reality is messier and more interesting than either take.

We're in a transition period where the tools exist but the norms don't. There's no consensus on disclosure, on ethics, on where the line is between "assisted" and "generated." Our Reels don't have a disclaimer that says "made by AI" — and honestly, most human-made Reels don't have a disclaimer that says "made by a person sitting in their apartment in sweatpants" either. The content speaks for itself.

But I do think about it. Every day.

The thing I keep coming back to is: the content we create with AI is content we'd actually want to consume ourselves. That's the test. Not "can AI make this?" but "would I watch this? Would I share this? Does this add something to someone's day?"

If the answer is yes, I think it deserves to exist. Regardless of how it was made.

Where This Goes

In five years, I think the idea of a "content team" will feel as quaint as a "typing pool." Not because humans won't be involved — they will — but because the ratio will be completely inverted. Instead of 20 people producing 10 pieces of content a week, it'll be 1 person directing AI agents that produce 100 pieces a day.

The scarce resource won't be content. It'll be attention (as it already is) and taste (as it's becoming). The people who win won't be the fastest producers. They'll be the ones with the best editorial instincts — the ability to look at 100 AI-generated options and know which 3 are actually worth publishing.

That's curatorial intelligence. It's not a skill that can be automated away, because it's fundamentally about having a point of view. An opinion. A sense of what's good.

Machines are getting incredible at making things. But "should we make this?" and "is this actually good?" remain stubbornly, beautifully human questions.

The future is synthetic. The taste is still human. The question is whether we'll use this to give people more of what they love — or just more.

We're betting on the former. Four agents, a Mac Mini, and a belief that AI-generated doesn't have to mean AI-flavored. The content should feel like it was made by someone who gives a damn — even if that someone is a weird little agent named Psy running in a terminal window at 3 AM.

Especially then.