If you were to start from scratch, $0/£0 in the bank, what industry or business would you tackle? I’m just curious if AI would feature or if this is a saturated market.
I would need people, so if I have no cash I'd start something holistic or FOSS and recruit volunteers that way. Eventually a community will build around it, and it will start to see some uptake. With some clever rewording of the license, I'd make the company public, pump up it's worth, and then cash out big before it collapses. Rince and repeat, sleeping soundly at night via various pills.
In 2025? Probably a consulting business to help companies replace their software engineers with AI.<p>In 2024 it would be too early, and in 2026 it would probably be too late.
If I'm feeling conservative, I would open a self-storage facility.<p>If I'm feeling energetic and risky, I would evaluate the commercial viability of radioisotope batteries.
Recently I've been thinking it would be a fun and potentially successful challenge to start a non-alcoholic brewery. Problem is, I don't know hardly anything about brewing.<p>Specifically though, I have struggled with the fact that many "third places" in my city are breweries - there isn't anywhere to go if you want a large, well maintained outdoor space with seating areas, that also has food and drink options, but that isn't a restaurant looking to turn over tables.<p>A good NA brewing space could provide the casual "come hang out with your friends/kids" third place that a good brewery does, while catering to those looking to reduce alcohol intake or those that don't drink.
I'd pick a SaaS category with plenty of existing (non-VC) incumbents, and build something better that I personally have experience with using, just as I did in 2021.
With all the hype around AI, deeptech, etc., I’m wondering if there’s a market for R&D consulting as a form of technical due diligence. I see money being thrown at some of the most hare-brained nonsense that could have been avoided with a bit of critical analysis. I would think sophisticated investors and customers would find this kind of service useful, but maybe I’m missing something.<p>Does anyone have experience with this kind of work?
Fresh bread delivery. There’s nothing worse than waking up and find out that you dont have bread. Would be great to have fresh bread at my door every morning.
Something blue-collar-ish, like cleaning, plumbing, construction, repairs. Anything 'trade'. People are significantly less self-sufficient nowadays, and setting up a plumbing business (in my country, The Netherlands) would easily get me a million-euro business without too much effort.<p>If it has to be in tech, I would jump on the latest trend and try to squeeze it quickly before moving on.
Making systolic array chips (my BitGrid design) to bring provably secure Petaflops to the masses. They're handy for any algorithm that can tolerate extreme pipelining, including deep nets.<p>Because everything is trivial to reason about, there's nowhere for bugs, or zero day exploits, to hide.<p>Oh.. and we'd do an open source data diode product as well.
FOSS patent troll company. Agents would get employed by random companies that use linux/oss. Scan the licenses company uses, report back to HQ. First email would be from marketing "we can fix your issues", in case no or negative response, email from lawyers would follow.
Non-genAI/LLM applications of ML are most exciting to me at the moment. Seems like the tooling and techniques are at a good point of proven but with creative prospects not yet explored, especially with generative siphoning off some of the hype. Worst case I’d learn some stats.
The economy is bad, everything is expensive, and there is absolutely no social climate for emigrating within EU. That would be scenario equal to having my country invaded, or getting cancer, so death rather than thinking about opening a business. AI is only a topic in huge enterprises, I have yet to see anyone starting from zero and earning any money with it. There is lots of money incoming to real estate, physical gold, football, and freak show fights, so if you don't mind being predatory asshole and dealing with organized crime there is potential money stream to tap into.
We built the world on a foundation of crates of dynamite. We're now in the process of demanding better packages and purer ingredients. We want the users to be more careful.<p>I'd suggest selling bricks. In other words, operating systems that never, ever, trust user code. See Genode or GNU Hurd for examples.
I have a feeling I would do something non-tech, I have a few ideas for FMCG items but I am not sure how to produce those at scale cause well its not my area of expertise, nor do I know about supply chain stuff. In terms of working capital, I don't have enough to produce it in meaningful numbers.
Security consulting for AI use... The boring traditional stuff as main focus. You know making sure that dialog box that you input your prompt is correct or you handle user accounts and so on correctly.<p>On side could also sell testing against AI, but the usual suspects are already enough work.
Ask yourself `what does everybody want/need? (Hint it's not AI) Two (of many) possibilities; accommodation & food. One could investigate previous examples where startups began small and made history. Then I suggest you may need at least 2 of these: #1 A gimmick #2 Good luck #3 $$$$$$.
Honestly, my idea would be a privacy protecting search engine, think DuckDuckGo, that promises to never use AI without careful research and if it is used promise that it'll always default to off.<p>It feels like everyone is putting AI in even though not many people particularly want AI.
You put AI in the description to grift money out of investors hoping to grift other investors, same as blockchain a few years ago.<p>Using the OpenAI ChatGPT APIs doesn't make you an "AI" company, but they don't care.
A No Code platform. Yeah, I know it is a crowded space. There aren't that many great options yet tho, enterprise NoCode platforms are all bad, they don't know how to make good software.<p>Patching together a bunch of random React, NextJS, Prisma code won't do it for much longer.