Since the launch of GPT-4 on March 14, the world has exploded with business ventures started from a simple ChatGPT prompt. Several founders shared their journey to test if this is a viable route to entrepreneurial stardom or if, indeed, it takes more than AI to build a business.
On the day of the launch, Greg Brockman, the president and co-founder of OpenAI, delivered a demo of their GPT-4 model, in which it could:
- Summarize an article in a single sentence where each word started with the same letter;
- Find common themes between two articles;
- Build a Discord bot on the spot;
- Read an image and describe what the image is about;
- Build a website from a handwritten note;
- Do a couple’s taxes.
It didn’t take long for people to make the leap and ask ChatGPT if it could come up with the next big thing in business. The next day, brand designer and writer Jackson Fall instructed ChatGPT to come up with a business idea that would turn a $100 investment into “as much money as possible in the shortest time possible.” The idea? GreenGadgetGuru, an affiliate marketing site focused on sustainable products.
Inspired by Fall’s efforts, the following week João Ferrão dos Santos — an entrepreneur and business mentor — put a different cap on the idea. His initial investment was $1,000 with one hour of human work per day. The AI immediately came up with an idea for a print-on-demand e-commerce store, AIsthetic Apparel, which sells T-shirts made from sustainable cotton and featuring AI designs.
We’re yet to see if the two ideas eventually flourish into actual, profitable businesses. However, the question remains: Can anyone come up with a successful business venture using AI?
It Takes More Than an Idea
Just because an idea is produced by a powerful algorithm that has access to arguably all of human knowledge, it doesn’t mean it will be a winner. After all, history abounds with examples of amazing business ideas that didn’t take off because they were ahead of their time, behind their time, or didn’t have the right leadership in place. The smart glasses, Google Glass, were launched in 2013 and despite unlimited resources behind the product and widespread adoption among celebrities such as Beyonce, Oprah, and (at the time) Prince Charles, the product flopped. In March 2023, Google discontinued its production.
Webvan, an online grocery delivery business, which was launched in 1996 and at its peak attracted $275 million in investment, declared bankruptcy in 2001 after a series of strategic mistakes by the leadership team who had no experience in the supermarket industry. The tech market crash in 2000 was the final nail in their entrepreneurial coffin.
Fast forward to today, and after I fed GPT-4 a similar prompt for a business idea, it told me to create an online course creation and sales platform. However, creating online courses is labor-intensive, so I specified ideas that don’t require a lot of upfront effort. The first answer was a print-on-demand merchandise store. The second idea? You guessed it — an affiliate marketing site.
Hallucinations and Contradictions
It’s no surprise that ChatGPT came up with the exact same business ideas as the ones it gave Fall and Santos. The tool is consulting the same database to give statistically probable answers. The prompts were also quite similar, prioritizing minimal human effort with a small upfront investment.
I asked ChatGPT to give me a specific affiliate marketing business venture, which led to TechTrekker, an online business focused “on promoting the latest outdoor and adventure technology gadgets, gear, and accessories.” Things got very interesting when I asked for specific market research data and market-gap analysis before September 2021 (otherwise the answers it gave were generic). The first link it shared, however, stated the very opposite of what ChatGPT was claiming.
It said that according to the Outdoor Industry Association, “the outdoor recreation market in the United States generated $887 billion in annual consumer spending” in 2019. But ChatGPT hallucinated that stat. The association itself has shared that “Americans went on one billion fewer outdoor outings in 2018 than they did in 2008” — contradicting ChatGPT’s claim that there’s an “increasing interest in outdoor activities among Americans.”
It could be that ChatGPT is referring here to a post-pandemic growth in interest, but that isn’t clear from its response. Had I taken AI’s lead here and started the business, I suspect I wouldn’t have gone too far. Before committing to any business idea, clearly we would need to conduct thorough research and dig deeper with more prompts to clarify any contradictions.
Automated Entrepreneurship Isn’t Automated at All
Two weeks after GreenGadgetGuru was launched, Jackson Fall generated over 20 million impressions, grew from 2,000 to over 100,000 followers on Twitter and started the HustleGPT movement on Discord. Not a small feat for any entrepreneur, but the business idea is yet to get off the ground. All revenue made so far has come from “investors” because Fall aggressively advertised his efforts on Twitter. No purchases have been made via the actual website because … well, there’s simply nothing to purchase yet.
AIsthetic Apparel, however, seems to be doing much better. Despite its limited number of designs, customers can place an order online and have their T-shirt delivered worldwide at the price of $38. The business sold out on its fifth day from launch date, shipping 122 orders to three continents in 24 hours. On March 31, Santos is hosting a 100-person event in Singapore on all things AI and business (all on the back of the virality of his LinkedIn post).
The outlook seems promising for AIsthethic Apparel, but the real test will be once the novelty wears off and the media attention dies down.
What’s evident from both examples is that it takes far more than a few prompts into an AI chat box to build a lucrative business. Both Fall and Santos have been relentless in sharing their experience online, responding and connecting to people, and using any media attention they can get to put the word out. Isn’t that what any founder does, though, with or without the assistance of AI?
While Santos signs off his LinkedIn posts as the “Executive Assistant to the CEO” (the CEO being GPT-4), it’s Santos who decides what prompts to feed the AI and who is building a business off his existing network. It’s no coincidence that he’s already an entrepreneur well-versed into go-to-market strategies and venture capital investments.
Should You Start an AI-Led Business?
If anyone is still wondering if they should follow AI’s lead in building a business, the answer would be — it depends. Are they already someone predisposed towards an entrepreneurial lifestyle who likes taking risks and is willing to put in the effort to get the idea off the ground? Do they have niche industry knowledge they can capitalize on? Are they ready to dedicate a good amount of time to the idea? If yes, then AI can be a great sounding board and ideation tool for new avenues for growth. If, however, they would rather just have the AI wire them the money into their bank account, perhaps they are already in the right job.
