The Uncomfortable Truth about AI and your Business in 2024.

The Bold Claim: If you ignore AI, it will have drastically negative implications on your business. If you adopt it correctly, you might create generational wealth. If you're an employee, that's riskier than being an entrepreneur now.

I believe this so strongly that I quit a lucrative machine learning job at FAANG to risk building my fintech AI startup. My software has led to full job automation of entry level employees, and I've seen my customers steamroll their competitors over the past year due to efficiency gains from AI. This is not a topic I want to talk about at all. In fact, it's not a topic anyone wants to talk about, which is precisely why I felt the need to write this. People in Silicon Valley get uneasy when job automation comes up and the most optimistic response you'll get is "We'll figure it out bro". People outside the industry are often met with the same uneasiness or feel completely confused – which often manifests itself as denial.

My goal then, is to explain why I believe my claim is true and create a discussion around how to take advantage of it.

…

_____________________________________

#1 – There's no indication that AI will slow down (a brief history + the trajectory):

• In 2017, I read a paper from Ilya Sutskever. The now former Chief Scientist of OpenAI. He co-authored this paper called "Attention is all you need". At a high level, the main idea is that with more data and more compute we can just make AI smarter. This scaling principle isn't just for text. It works for images, video, music, voices, vision, and evidently robotics.
• Shortly after the paper was released, GPT1 was released. Then GPT2 a year later. Then GPT3 the year after that. Every time a new model was trained, it has proven that if more compute and more data is thrown at these models, it just scaled and became smarter. There are emergent properties that researchers never predicted – and more importantly there is no sign that this will asymptote or slow down. Hence the rise of NVDA stock.
• IMPORTANT*: This is why venture capital dollars are in the hundreds of billions. If it can keep proving this is the case, then more money will flow for the next iteration to be better. The goal of these research labs is very clear. They want to create AI that surpasses human intelligence. Depending on how smart it is, the first company to do this could hypothetically form a global monopoly and change markets as we know it. I do not think this will happen soon, but it's crucial to understand that this feedback mechanism between* more $ = Smarter AI means that we could be on an exponential trajectory. The dangerous thing about exponentials is that by the time it takes off, you were either "lucky" and early or too late to react because it all takes off at once.
• So right now, someone may say "GPT isn't smart enough to take my job", but in previously models it couldn't even form a sentence or reason. These labs are actively working on giving these systems the ability to work on tasks that would take a human several months.

…

_____________________________________

#2 – AI Agents are going to be the new employees (With examples)

That's not to say employees will be eliminated. Certainly not, but many jobs will be. AI is slowly cannibalizing the corporate ladder which will make the job market more competitive. AI labs like OpenAI are focused on creating what are known as "foundation models". They spend billions of dollars on training sophisticated AI systems like GPT and they either open source them or create an API for startups to use. The startups that are actually doing well are known as "agents". They use a foundation model along with traditional software engineering to create an AI that can do a specific workflow. Right now, these are limited mainly by the capability of the underlying models. My business is a form of an AI agent, and as I mentioned, its already having drastic impacts on my customers. I'll give a few examples of the most common that are mentioned for general purpose use cases:

_

AI doing Sales – SDR/BDR:

• This is likely the most famous agent at the moment. It's an AI agent called Mindy, that you just email. It was co-founded by the founding team behind PayPal and YouTube along with an executive from Accenture. The AI is simply something which you email. It can answer questions like ChatGPT, read documents, but most interestingly, scrape the interest and give you information from websites. People are primarily using this to automate their prospecting and outreach process, similar to what an SDR would do. Users just email the AI what kind of customer they're looking for (company size, industry, location, etc.) and it will scrape the Internet and return a list of all relevant names and companies with personal information and auto draft emails to them. FWIW, this is currently free but I'm unsure of their future monetization plans.

_

AI Programmer:

• Probably the most ironic out of all of them as you'd think the last thing AI would be able to do is program. This is not replacing humans yet, but I'm including it because the potential is incredibly close and will have a large impact. There's a company called Cognition founded by a team of Harvard math Olympiads that is automating the software development process. It has access to libraries, a terminal, and an editor. Granted, this is mostly used to assist developers currently. But what's notable here is that it's only limited by the underlying language model. Meaning, they've built out the entire skeleton for when the next model comes out. If the leap from GPT4 to GPT5 is as great as GPT3 to GPT4 on programming tasks, it would undoubtedly eliminate many junior and entry developer roles. Managers and senior devs would simply review AI written code.

_

AI Phone Calls:

• AI Phone Calls are also automating jobs already. The primary company behind this is sending out hundreds of thousands of calls a day. Is it able to have a full-on conversation for 40 minutes? No. But it's doing T1 customer support, lead qualification/scheduling, and automating plenty of internal operations where employees were previously making phone calls.

_

NOTE\: The kicker here is that all of these are a fraction of the cost of a human and work 24/7. These are just a small number of what exists out there. I have friends that have setup entire automation flows around their company that they would have previously employed 10+ people to do, and all of it is happening automatically. One can easily see why entry level roles are at risk. If the models improve, higher up positions become at risk as well. Fewer middle managers will be needed since there's fewer employees. This means there's less employees that produce more output. The good thing is that enterprises are slow to adapt. People won't trust agents for anything that's not a simple task for a while which gives smaller companies time to take larger risks.*

…

_____________________________________

#3 – Closing Thoughts

Again, it's never been a better time to be an entrepreneur. It's when massive societal changes occur that the real winners come out and create life changing wealth. The risk of not doing your own thing is more than just regret, it's that you're essentially betting that the AIs won't get better. If you already have a business and don't adopt AI, there are people that understand that it will get better. They're staying on the cutting edge. The best thing everyone can do is:

1. Find AI tools that can increase your revenue and also find tools that are vertical specific and give you some kind of edge. For me this would be AI dev tools. If you run an ecommerce store this might be using AI for product recommendations. If you run a law firm, maybe it's buying an AI that extracts key pieces of information out of legal documents and provides citations. It doesn't matter what business you run; I guarantee you that AI will impact it.
2. Stay on top of the news. You're already doing it by being here but go subscribe to AI YouTube channels, podcasts, and newsletters (The Neuron, TLDR, to name a few). Follow pages on social media. This is bigger than the Internet, and we all know what happened to businesses that refused to create websites and use email. There are probably a dozen tools that could lead to you increasing your revenue and efficiency by massive margins.

If after all this, you still think AI won't impact you because ChatGPT is "meh" right now, then I say good luck soldier. Things are about to get wild.