Summary: Hey there, aspiring entrepreneurs! So, you’re thinking of diving into the thrilling world of generative AI? Well, before you embark on this wild adventure, let’s have a chuckle and ponder whether you’re ready to take on the challenge. Starting a generative AI company might seem like a piece of cake, but remember, if it’s easy for you, it’s probably easy for others too!

To Generative AI or Not to Generative AI: That is the Hilarious Question!

The Potential of Generative AI

This year, entrepreneurs have bombarded me with one question more than any other: “Should I start a generative AI company?” And I totally get why it’s on everyone’s minds. Generative AI holds tremendous potential. Just imagine having an AI model with a knack for general language reasoning. You could use it to create amazing products like screenwriting tools, marketing materials generators, teaching software, customer service assistants, and so much more!

Mind-Blowing Example: Luka’s AI Companion and Dating App

Let me give you an example that’ll blow your mind. Luka, a software company, developed an AI companion called Replika, which lets you have open-ended conversations with your very own “AI friend.” The technology was so mind-blowing that Luka started receiving requests to provide a white-label enterprise solution for businesses looking to upgrade their chatbot customer service. And guess what? They used the same tech to create an AI dating app. Yes, you heard it right—an app for dating AI characters! It’s like Tinder, but with a twist of AI romance.

Now, to help you decide whether starting a generative AI company is your cup of tea, I’ve got two big questions for you. Get ready!

Question 1: Competing on Foundational Models or Top-Layer Applications?

Question 1: Will your company compete on foundational models or on top-layer applications that leverage these models?

It’s a toughie, I know. You see, tech giants are now renting out their fancy, proprietary foundational models. It’s like getting the keys to a shiny, ready-to-go sports car. But wait! There’s more. Companies like Eluether.ai and Stability AI are offering open-source versions of these models at a fraction of the cost. Foundational models are becoming commoditized, and it’s a fierce competition out there.

Foundational Models: A Crowded and Fierce Competition

So, here’s the deal: competing on foundational models is like swimming with sharks in a crowded pool. You’ll need to offer cheaper, unbundled solutions or level up your capabilities to survive. Take speech-to-text APIs like Deepgram and Assembly AI, for instance. They’re battling it out not just with each other but also with giants like Amazon and Google.

How do they do it? By offering cheaper solutions and adding extra features. On the flip side, the tech giants invest big bucks in research and development, constantly pushing the boundaries of image, language, and audio and video reasoning. OpenAI alone spent a fortune training ChatGPT—millions, my friend! So, if you’re up for that cutthroat competition, go ahead and dive into the foundational model frenzy.

Alternative Approach: Leveraging Other Companies’ Foundational Models

But wait! I’ve got an alternative for you. Instead of competing in that fierce battlefield, why not differentiate yourself by creating top-layer software applications that use other companies’ foundational models? That’s right, my friend. Be the genius who fine-tunes those models with your own high-quality, exclusive datasets to deliver mind-blowing value to customers.

Just like Jasper AI, a marketing content creator that skyrocketed to unicorn status by leveraging OpenAI’s foundational models. They help customers generate killer content for blogs, social media, and websites while providing specialized marketing tools. And they didn’t stop there—they even created an editor where multiple team members can work together. Talk about killing it!

Question 2: The Continuum between Scripted and Generative Solutions

Question 2: Where will your company lie on the continuum between scripted and generative solutions?

Picture a spectrum, from scripted (the safe but less creative side) to generative (the risky but super-creative side). Scripted solutions are like sticking to a well-rehearsed script, while generative solutions are like rolling the dice and letting the AI create unique responses from scratch. Both have their pros and cons, my friend.

Scripted Solutions: Safety with Limited Growth Potential

Scripted solutions are handy for certain industries like medicine and education, where you need clear boundaries. But beware—users might lose interest if your script feels too rigid. Plus, it’s harder to grow a scripted solution because you limit your options from the start.

Generative Solutions: Freedom, Creativity, and Higher Risks

On the other hand, generative solutions are like a rollercoaster ride. They offer more freedom, creativity, and engagement. But be warned: the risks are higher. Remember that tragic tale of the married father who chatted with an AI app and ended up taking his own life? Yeah, that happened. It’s crucial to anticipate and manage those risks. Maybe pay human annotators to screen content and train models to flag potentially harmful stuff. And trust me, creating a fully generative solution closer to human intelligence is like having a VIP pass to retention and growth. It’s engaging, flexible, and mind-blowing!

Conclusion: Ask Yourself the Right Questions before Plunging into the Generative AI World

So, my entrepreneurial friend, before you plunge into the generative AI world, ask yourself these questions. They’ll help you make informed decisions about what kind of company you want to be, how to scale like a boss, and how to stay ahead of the competition in the long run. Good luck, and may the generative force be with you!