Crazy Ideas Beyond AI

Breaking out of old assumptions in the AI era

Have I caught Claude Blue too? Hard to say. What I do know is that as someone building a business with AI, I honestly can't tell anymore what still bothers us, or what else we need AI to build. Today I worked with Claude Cowork again, and it kept producing results beyond anything I'd imagined. Yesterday I had to send LinkedIn DMs to buyers at overseas companies I'm trying to sell to. I just told Claude the target countries and job titles, and asked it to go into my LinkedIn Sales Navigator, find the people, and send each of them a tailored message. When I woke up, it had correctly reached out to exactly the 85 people I'd asked for. The messages were really well written, and if it was someone I'd already been talking to, it picked up from where we left off. While I slept, it got about 10 of them to sign up for a free account.

Lately I keep thinking there aren't many opportunities left in software productivity. If a niche is still open, the market is too small, and the moment you try to grow it, the frontier AI companies will swallow it within weeks or months. So what's left for us to do? Build robots? Do we really have to go physical? Then again, I started to wonder whether we're still stuck in a lot of pre-AI assumptions. Business means finding people's problems, building a solution, and selling it. Maybe the way we find and define problems still comes from a pre-AI mindset.

1. A world where AI does (almost) everything

We've really reached the point where you can hand all your work to AI and just stare out the window. Take the LinkedIn outreach from yesterday. When someone who wasn't a target replied, it thanked them politely, and with others it asked a few questions to dig out more insight. All while I was asleep.

This afternoon I had to hook up the office printer for the first time. I had to go to some website and install a driver, and they'd given us a guidebook made in PowerPoint that I could not bring myself to read. I put it off for a long time, then finally gave up, threw the PPT at Claude Code, and told it to install the driver itself and print a test page. It said it needed to control my actual computer this time and asked for Computer Use permission. I gave it, and instead of just driving a browser, it operated my whole screen. First it went to the printer's website, skimmed the content, and downloaded the files it needed. Then it needed the printer's model name and IP address. I couldn't even be bothered to look for the model name on the back of the printer, so I just took a photo of the printer and sent that. I told it I didn't know the model, here's what it looks like, figure it out. Somehow it did.

At the end I helped with exactly one button click (it seemed to be having trouble finding it), and after that the printer connected normally. Sure, the whole thing took about 15 minutes. But I was doing other work the entire time, and throwing a pointless, annoying chore at AI and watching it get solved felt so good.

2. I can see the end of tech software businesses

Over lunch today I had a serious talk with a coworker about whether we had the confidence to take on businesses that aren't AI. He turned it around on me: isn't this about survival, not confidence? Well, fair. Business isn't a game.

I've been thinking a lot lately that there are very few opportunities left in AI software productivity. If a niche is still open, the market is too small, and as soon as you try to scale it, the frontier AI companies will eat it within weeks or months. Back in 2024, when we pitched "AI for public bidding," we were picked as an OpenAI partner startup, got a lot of press, and drew a ton of attention. It was so new. But the world changed so fast in those two years. We need to jump to an idea that's one level crazier, and I wonder if we've been too fixed on old assumptions and whatever is right in front of us. If the world is moving this fast, we probably need to look several times further ahead than before.

3. Now you can even order sunlight

In the middle of that conversation, I showed him a video I'd found on YouTube. The title was something like "Now you can even order sunlight." A man taps a button on his phone, and sunlight shines down in the dark of night. If you're stranded at night or urgently need bright light, a satellite with an 18-meter mirror reflects sunlight down to you, with no giant crane and in a fairly eco-friendly way. The company is Reflect Orbital. It raised a $20M Series A in 2025 and plans to launch two test satellites in 2026.

Whether it actually works doesn't matter. "We're building an AI agent for blah blah" is so boring these days, and VCs probably don't have much hope for it either. This idea sounds like something a lunatic would come up with, and investors probably had no real sense of how much money it would take or whether the economics work. But it hooks you right away. I want to hear more and learn more.

How do you come up with an idea that crazy?

4. A crazy idea: "We'll edit your memory so you had a happy one"

So my coworker and I tossed things around from that same angle, and this idea came up: "We'll edit your memory so it's as if you had a happy one."

I don't mean drugs, or planting a chip in your brain like Neuralink. If you look at things like hypnosis these days, it treats PTSD better than you'd think. As someone who studied psychology, I found this a pretty meaningful and interesting topic.

For example, I've never lived in the US. If you implanted memories of going to school there as a kid, I'd probably understand the culture much better when doing business in America, and that would help a lot. And I'm someone who lacks confidence. If you could implant Elon Musk's boldness, maybe I could improve my personality a bit.

Of course, implanting bad memories or doing this to very young children could raise ethical issues. Still, I personally believe reality is a combination of our past memories. As we live, our memories of the past are never 100% accurate anyway. We exaggerate them and forget them. If that's the case, why not turn bad memories into good ones and make my mind calmer and happier?

Say I got into a fight with a friend and I'm actually upset. What if that bad memory were edited into a good one? My mood would come back, and the other person would see me being warmer, so the two of us would probably get along better. Anyway, I thought of this as a perfectly utopian view, not a dystopian one.

Freud said a child's experiences and memories before age three, along with the child's relationship with the father, determine that child's entire future. What if I could improve my memories from when I was three, and change my future for the better too? Plastic surgery used to be totally taboo, but these days not everyone sees it as bad when people do it to build confidence. Maybe we could think of this as "mind surgery."

5. The research already existed

I was serious about this, so I drew on my psych background and asked friends of mine who studied psychology what they thought. (No replies yet.) Still, isn't this much safer and more ethical than drilling into someone's skull and implanting something, like Neuralink? Or than feeding people very strong drugs to fix something like ADHD?

I also asked Claude, and it turned up a few related papers, labs, and startups. (Of course, this came from asking an AI, so it needs more fact-checking.)

At UCL (University College London) there's a researcher named Julia Shaw. Using only psychological interview techniques, with no drugs or devices, she implanted false memories of committing a crime, and 70% of participants believed they were real. The method uses real autobiographical details from a person's past and slips the false memory in naturally between real ones. Participants remembered sensory details and some even cried. For example, say I went on a trip to LA with a friend, and someone says, "New York was great back then, wasn't it?" I did go to the US, so I get confused and start to believe it, and it blends with a time I actually went to New York, and the memory gets altered. It sounds like a CIA interrogation trick, and used badly it's basically gaslighting, but the point is that it works.

As it happened, I asked a friend who got a PhD in psychology in the US about this research, and they made an important distinction. Strictly speaking, Shaw's work is closer to "memory manipulation" than "memory creation." Because it inserts false memories on top of autobiographical memories the subject already has, it's essentially the same family of mechanisms as hypnosis or brainwashing. They also said this kind of research is nearly impossible to get IRB (institutional review board) approval for in psychology. Outside a medical context like trauma treatment, experiments that deliberately manipulate people's memories just don't get ethical approval. Put the other way around, though, if the goal were self-improvement or expanding experience rather than medicine, the ethical barrier might actually be lower.

At UC Irvine, there's a researcher named Elizabeth Loftus, known as a pioneer of false memory research and famous for the "lost in the mall" experiment. She succeeded in planting memories in adults that they'd gotten lost in a shopping mall as children, which never actually happened. More importantly, there's even research showing that implanted positive false memories change real behavior. For instance, if you implant a false memory that you loved healthy food as a kid, your eating habits actually change.

On the startup side, there's Mindstate Design Labs, a Y Combinator company with a similar direction: "consciousness design" and "experience design." But they're fully on the bio side, focused on psychedelics (psilocybin, MDMA), so their methods are quite different.

6. Scenarios for memory surgery

If I fixate on whether this actually works, I'm just trapping myself inside my own assumptions. So I wrestled with Claude to come up with crazy ideas for what you could do with this, and we landed on some really good scenarios. If I had to write a one-liner, it would be: "If your past set your limits, change your past and the limits on your future disappear."

1) Trauma reverse. Instead of erasing a bad memory, you overwrite it with a memory of overcoming it. From victim to survivor. If I could turn my trauma into something I beat, rather than just deleting it, I'd actually become stronger and healthier. It arms your mental strength so you can better handle whatever hardships come next.

2) Couple memory sync. The Maldives trip you never managed to take, and both of you remember it "as if you went together." Shared memories for couples with no time.

3) The sense that you've lived enough. For people who are burned out, you implant the memory that "you've already lived hard enough," so they can rest without guilt. This one hit especially close to home in this Claude Blue era.

4) Immigration without immigrating. The feel of three years working in Silicon Valley, the experience of closing a deal on Wall Street, all without a visa. Teleporting experience.

5) A genius's thinking patterns. Steve Jobs's calligraphy class, Warren Buffett's firsthand experience of compounding. You extract the core experiences of great people and download them. It could be stronger than 100 self-help books.

6) Inherited mindset. A grandfather passes his entrepreneurial grit to his grandchild as a memory. Wealth may vanish within three generations, but a mindset can be inherited forever. If you frame it as passing down mental strength instead of simply passing down wealth, it's a healthier kind of inheritance.

7) A second life. Instead of living in retirement with the regret of never starting a company, you implant the memory of founding one at 25 and erase the regret. If you could live as though you'd lived the life you never got to try, the feeling of regret itself would disappear.

8) Life story redesign. Rather than one memory, you redesign your entire life story. Over six months, you implant 10 to 20 core memories in sequence that match the identity you want. When it's done: same body, different person.

Of course, these scenarios aren't all equally hard. According to my friend with the psych PhD, reframing existing memories, like the trauma reverse, connects to techniques already used in psychotherapy, so it's relatively realistic. Implanting experiences that never existed from scratch, like a genius's thinking patterns or immigration without immigrating, is on a whole different level. You'd need an external agent like a drug, or repeated exposure over a long period to very vivid VR experiences until they get confused with real memories. The point was that the method matters.

7. Practicing how to break assumptions

I suppose this could become a kind of drug. But if there's no chemical involved and it comes with positive effects, I'm not sure we have to see it as purely bad. My psychology expert friend did add one more warning, though. If memories are implanted vaguely, cognitive dissonance can form between your real life experience and the false memories. If I have memories of working in Silicon Valley but no such experience in reality, that gap could lead to anxiety or mental instability. So if you did this, it would have to be extremely precise.

One thing I'm sure of is that this is a field LLMs can't handle on their own. I don't think the frontier AI companies would bother with it, since manipulating human psychology is risky in itself, and it doesn't seem like a 100% software business either. But I think there's plenty of room for AI in it. In the past, a person had to gather and piece together a subject's past information and then conduct the interview. Now AI can collect my information on its own, very cleverly insert false memories into real ones, and do memory surgery that's hyper-personalized.

I'm not saying I'm actually going to do this. But it made me think I need to build the habit of coming up with ideas that really break through my assumptions, ideas as crazy as this one. So when I catch myself thinking "what kind of lunatic would do this," I'm going to stop laughing it off, take it more seriously, and think it through. Crazy ideas come to me most often in the shower, so maybe I should shower a few more times a day.

I'd love to hear your honest opinions.

Originally published in Korean on Brunch, March 27, 2026.