My first (failed) business

The year was 2023. I just moved to Sydney, and the company I've been working for had clear signs of instability.

Now, instead of looking for a new job, I wanted to dip my toes into the world of entrepreneurship. My reason back then was simple. I wanted to FatFIRE. I wanted to become financially independent with the freedom to spend on whatever I want.

Few days later, Luis Flores, a good friend of mine, sent me this one video. It was a guide to starting an AI Automation Agency (AAA). Perfect timing, indeed.

What is an AAA?

Think of it as some service you provide to other businesses to automate their processes and tasks using AI.

Basically, owners of AAA build custom AI solutions based on client needs. Your customers usually go to you to:

• Automate repetitive tasks

• Optimize decision-making processes

• Enhance customer service

• etc.

But wait, countless agencies have been solving those problems for ages. What changed?

LLMs.

Take the problem of enhancing customer service.

There are loads of chatbots out there. But, an LLM-powered one is unparalleled. They can act as human as possible, and, most importantly, understand context extremely well. Gone are the days when you have to click buttons to navigate a chatbot just to find an answer. With LLMs, just type your question, then you receive an answer.

How about a much harder problem like optimizing decision-making processes?

Let's say the client takes too long to decide which part of their product needs to be improved based on customer feedback. Well, easy. Feed all those written feedback from customers into an LLM, then have it summarize the main points. Let's take it further. Have an internal chatbot for the business so the stakeholders can ask free-form questions about those feedback. From raw data to insights in less than a minute.

It's clear. Because of the rise of LLMs, solving typical automation problems in businesses became accessible. Not to mention the countless no-code tools out there to help you build these technologies.

I was sold. I had to start this.

Selecting my Niche

No question about this, narrowing down the niche allows AAA owners to focus on common problems of clients within that market.

I used Alex Hormozi's book $100M Offers to narrow down my niche based on the following guidelines:

• Purchasing Power: can a customer pay for your service?

• Targeting Ease: can you access the customer easily?

• Market Size: do you have lots of potential customers?

• Familiarity: do you know the industry?

• Automation Potential: do the customers have problems solvable by automation?

It was a fairly simple guideline. So, I listed random markets, scored them, then arrived at the following result.

Not much thought went into tie-breaking. I just had to get started, so I chose a random niche I was interested in: Food Delivery Services in the Philippines.

I also did more market research as initial validation. And oh boy, was I surprised! Their top player has $41M in valuation, while #2 has $1.1M in ARR. Crazy numbers.

A pretty good niche, if you think about it, except for one thing. The market size. You’ll see in the 2nd part of this blog why this bit me in the bum.

Offer Creation

Okay, now I had a niche. Next step was to think about what to sell. An offer.

I started with the value equation from Alex Hormozi. It has a neat formula:

where:

• Dream Outcome: what the customer wants

• Perceived Likelihood of Achievement: the probability of getting what the customer wants

• Time Delay: how long 'till the customer gets what they want

• Effort and Sacrifice: how difficult it is to get what the customer wants

Now it's simple math. You increase value either by increasing the numerator or decreasing the denominator. Easy, right? Ayt, let’s move on to each variable.

Let’s start with the dream outcome. I ended up with the following:

I was attempting to solve customer support issues of food delivery services through a customer support chatbot.

But here's how things started to go wrong. I never validated whether my customers had this problem or not. They say: “Don’t marry your idea”. Well, I married the niche, problem, and solution all at the same time. I didn’t have the mindset to treat this as a mere experiment that needs validation!

As inexperienced as I was, I continued spending days perfecting my offer when I should’ve just iterated quickly. But, let’s continue.

Perceived likelihood of achievement seemed too difficult to measure. I simply listed down numbers and case studies to persuade customers why they need us.

As for time delay, assuming I was a one-man team, I could deliver an initial chatbot within 24 hours. All thanks, of course, to no-code chatbot tools out there with LLM support.

Lastly, effort and sacrifice is close to 0. The customer only has to help me flesh out the chatbot flow.

Putting the value equation variables all together, plus some marketing tricks (e.g. urgency, scarcity), I ended up with my final offer:

Retrospective

It wasn't obvious at first. But, one of my biggest regrets when starting this business was I never fully understood what “Fail fast” meant. I was fixated on my niche, problem, and offer that, when I failed to acquire customers - I was literally lost.

However, I would still 100% recommend the approach from the $100M Offers book. In the book, there is a reason why the reader is prompted to select multiple niches, list down multiple problems, and create multiple offers. All of them are valid options.

You just need to go out there. Talk to customers. And find out which niche x problem x offer works (don’t ask me how tho, I still haven’t reached this far).

In the next part, you'll learn more about my mistakes building before validating, generating leads, and reaching out cold. Stay tuned!