Skip to main content

As someone who is highly utilitarian, I turn to Instagram to educate me on which colored walls would look best on my plain drywall, San Francisco home. I don’t know how to articulate what I want because I don’t know what I want, but usually Instagram will show me a photo out of the thousands, that makes my desire actionable—that’s my process.

So it was a surprise to me, that when I habitually tapped the search icon on IG, all of a sudden the flow had changed. What had once been images of accounts that I might be interested in, whether because my friends had followed these accounts or because Instagram knew I was going through a small renovation,  was now replaced by a clean chatbot interface and example search prompts.

Who's Training Who?

I did a double-take—was this my search thread? That couldn’t be. Hesitantly, I typed in “stylish windowless bathroom,” a prompt that had always netted me images of influencer’s fancy moody bathrooms, but this time came a rolling text scroll similar to my prompt with ChatGPT.

This wasn’t what I was looking for. Somehow, Instagram had thrown me into a new product flow. Where I was expecting images, I was now seeing a text roll and was presumably expected to converse with a model on what defined a “stylish windowless bathroom.”

It took me a second before I realized that Instagram’s new Llama 3 model was a hammer, and they thought I was a nail. Taking the bait, I commanded Llama 3 to “send me images of a small bathroom with black trim and moody blue walls, no windows.”

send me images of a small bathroom with black trim and moody blue walls, no windows.
Nice.

BAM. In what seemed like 3 impressive seconds, Llama 3 delivered a self-generated image (I checked the watermark) of a stylish bathroom.

Impressive.

The image was not perfect, so I asked it to remove the wainscoting and to remove the window it gave me, and in another 5 seconds, I generated another image for my inspiration.

Stay in-the-know on all things product management including trends, how-tos, and insights - delivered right to your inbox.

Stay in-the-know on all things product management including trends, how-tos, and insights - delivered right to your inbox.

  • By submitting this form, you agree to receive our newsletter and occasional emails related to The Product Manager. You can unsubscribe at any time. For more details, please review our Privacy Policy. We're protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
  • This field is for validation purposes and should be left unchanged.

Great for Users, Very Bad for Entrepreneurs

As someone who has tinkered with a few of the models: Bard, Claude, Chat GPT, Midjourney—this was a surprisingly quick, and even more surprisingly usable, rendering of an image.

Excited, I saved the image to my phone and showed my husband how I planned to update our home. It should be noted that I am highly particular about design despite not being particularly adept at it.

I could only imagine if you were the founder of one of the many remodeling AI apps on the market, you’d be thinking to yourself, “Oh shit,” which is reminiscent of the days when Pinterest and then-Twitter were building out their ecosystems, eating up market share of the smaller feature startups. (In fact, I remember many Twitter posts with the verbiage “oh shit” every time Twitter came out with some new feature.) This was more than a decade ago in San Francisco, but one can see that we are in a very similar, albeit faster-paced, land grab.

I also reside in Facebook groups for crowdsourcing design ideas for the above scenario, so I have first-hand witness to those using and paying to use these AI apps to do exactly what I did with Meta’s Llama 3. Users will pay the apps a monthly subscription that averages around $99/year to generate photos of their homes but in different designs and colors. However, in my case, I did it for free.

That led me to thinking, I’m sure there are a significant amount of enterprising individuals and even product managers who are thirsty to capitalize on the AI space—but does that mean you should right now?

So hear me out..

Maybe you shouldn’t quit your day job for your startup idea just yet...

...for the following three reasons.

  1. The job market. It's tough out there. If you quit something that you have experience in, it might not be easy to get back on the horse.
  2. The competition isn't just aggressive—it's murderous. AI is a hammer, and the companies with money are looking for nails; your little niche could be eaten up with the next model.
  3. Your (human) opponents are smarter than ever. - PhDs and people who normally do not compete in the industry space have been brain drained from universities and are now in industry. That means that there is an added element of competition that people in industry typically haven’t had to compete with.

Reason #1: The Job Market

Let’s start with the obvious one we can all agree with: layoffs are rampant, interest rates and inflation are super high, companies are posting fake jobs and ghosting candidates - even highly experienced people are taking a year or longer to find jobs in certain industries.

To put it quite frankly, this is no great resignation; and we the people do not have as much leverage as we did during the pandemic. Suppose you have managed to evade layoffs for your company so far. In that case, it might be wise to keep that steady paycheck and strong benefits, especially during a time when incredibly experienced candidates are taking lower-paying and lower-level jobs to pay the (very-high-inflation-impacted) bills.

Instead, try and work on your startup idea as a side hustle, if you are able to create bandwidth and also able to not impact your day job performance. I know not everyone is able to do this, whether having family responsibilities or having an all encompassing job.

All I’m saying is that we can agree that right now is not an easy time to take a risk letting go of a stable job that you’re really efficient at, especially if your backup plan is to get rehired right away if things don’t pan out.

Reason #2: Lethal Competition

This next point is a real hot take. Companies are voraciously hunting for consumer and commercial use cases for AI. They can do that given they have all of the financial resources to purchase GPUs and train models. Perhaps the initial versions of the models were incomplete and clunky, lending an opportunity for enterprising individuals to create layers on top of these models for personalization into their areas of expertise such as Legal counsel, Home Remodeling, Video Editing, or even Presentations.

However, as more money enters the AI space, there exists more money to buy more GPUs, and to retrain models to become more advanced. With each subsequent model and its improved capability, there is more opportunity to deliver a near similar value proposition to the end user for free. Like I said, in one fell swoop,  Meta’s most recent Llama 3 model was able to deliver an image to me for free that someone else paid $8 per month for. 

Reason #3: A Business Degree Won't Cut It Anymore

The last point is one based on competition. Universities have experienced a pretty significant brain drain from wealthy tech companies filled to the brim with cash, poaching their AI specialized PhDs and professors out of academia and into Industry. How this impacts the lay person, is that you are now competing with a highly specialized, highly knowledgeable person who typically you wouldn’t be competing head on with; they would usually be focused on finding funding to prove their scientific hypothesis in the name of science.

Shown: The MIT scholars your competitors just hired. (Source)

Now, you’re competing with them head on in the name of profit, as they have been lured out of the shadows of academia and into the industrial sunlight. You might know your user a bit better, but they have the foresight of their technical knowledge in what is right now a technical race. Whatever optimizations they create with their knowledge, goes straight into the wallets of the giant companies that employ them.

Do you really know more than these PhDs and professors who have invested in their pursuit of Artificial Intelligence for decades?

My prediction on how this last hot take manifests is through one of the most common questions from VCs to entrepreneurs: What is your moat?

Sure, you’ve figured out this one niche has a problem and you can solve it. But how do you defend your business if Meta or Google decides to pivot one day, and take over your market share with all of their resources?

If you are building out a solution that requires a technical understanding and the knowledge of foresight on where the market will go, you’ll need to address that the big tech companies have incredibly knowledgeable AI scientists who are leading the charge and not necessarily sharing that information with the public given that they are now employed.

Caveats: You Might Still Have a Fighting Chance

I don’t want to deter entrepreneurs with conviction who have a strong vision and call to action to build the future world now that AI is available. The following are reasons as to why you could absolute win in this new land grab for market share:

  1. If you have a very specific skill set that sets you apart from the others (ex: trained knowledge and high accuracy in an area of medicine or law) that allows you to be able to set very accurate prompts with high accuracy results
  2. You can compete technically with these PhDs on the technical front because you have knowledge of how these technical models work
  3. You are a statistical thinker and are ready to pivot when every single model you build your product on updates to the bigger, newer model.

Ultimately, the elephant in the room that should be addressed is that Open AI, Anthropic, Meta, Google, Microsoft, and all these companies investing in the resourcing to build these AI models are doing just that—building an all-powerful, all-capable model that right now doesn’t have a particular focus.

The argument is that they are looking for enterprising entrepreneurs to help them find the next trillion-dollar application and end-user use cases to build a successful company, similar to the financial model of Cloud computing and AWS.

You and your co-founder pitching VCs. (Source)

This is a fair argument, and one that ultimately solidifies why this whole thing you're reading is a hot take.

My point, that I’ll end with, is that unless you are highly specialized in your particular area of expertise—or just plain ready to work your ass off because entrepreneurship is not easy to begin with and this is entrepreneurship on Impossible mode—we just don’t know where things will consolidate from here.

It might make sense to observe and be smart, and build a side hustle without quitting your day job for now.

In the meantime, I’ll be curiously generating fantastical ombre hallways that likely no designer would ever agree to execute on.

Don't forget to subscribe to our newsletter for more product management resources and guides, plus the latest podcasts, interviews, and other insights from industry leaders and experts.

Victoria Ku

Enfant Terrible, Clandestine Artist, and Reformed Capitalist turned Product Leader and maverick. Victoria has spanned multiple industries, finally landing in tech where she spent 8 years at Airbnb launching a myriad of disruptive products (Airbnb 4 Work, Airbnb 4 Real Estate, Cohosting, Magical Trips etc.) before leading global payments platform strategy. Today she is at Highnote leading Product and Design, continuing to fight the good fight in fintech for innovation, efficiency, and inclusivity. In her free time Victoria sculpts, is an avid reader, and tends to be a contrarian; she insisted on having an Aliens versus Predator themed wedding.