Here’s the back story. In 2023 I weighed over 320 lbs and I was desperate to cut the weight. I locked in on the gym and cleaned up my diet. As part of cleaning up my diet, I needed to keep track of my calories and nutrients. There were many tools and apps available to do that, but they all cost money. Cal.ai costs $10 a month and MyFitnessPal costs $19.99 a month. I didn’t see a need to pay that much for an app that tells me how many calories and macros are in my food. At that time I had just finished building Scriblo, the app you’re reading this article on right now 😅, so I had a lot of bandwidth to take up another project. I set out to build my own calorie and nutrient tracking application and I called it Calquest.

I wrote the first line of code for this app in 2023 and I slowly continued to develop it on an on and off basis until it was finally shippable in December 2024. Even then I still had every solo founder’s fear: “I don’t think it’s ready for users.” That mindset pushed the beta release to August of 2025, almost a year later. In August 2025 I released the beta version of Calquest to 100 plus users who signed up on the waitlist. There was a lot of excitement around the app and the waitlist saw close to 200 signups. After onboarding the first 100 users, that’s when reality started to set in that “I might have actually fucked up.” And this was something I would have caught if I had allowed my dumb self to release a lot earlier.

A very brief info on what Calquest actually is for context:
Calquest had 3 main functions. It helped users track their calories and macros, helped users who were trying to lose weight get into intermittent fasting, and tracked their fluid intake. Those were the 3 main building blocks of Calquest. For the most part it did those 3 things almost flawlessly. When it came to tracking calories and macros it did that perfectly. It let you look up literally any food that exists and scan the barcode of any packaged food to instantly get the calorie and macros information. You could then log based on how many servings you had. You didn’t have to do any math, just scan and log. Tracking your water intake was also simple and intuitive, same with intermittent fasting. One other key aspect of dieting is being consistent, so I thought about “how can I make people want to log their meals every single day?” The obvious answer was notifications, but notifications in similar apps were too generic. There was no motivation to even click on them. This pushed me to design probably the best notification system I’ve ever seen in any app. You can read more about how I developed a personalized notification system . The whole mantra behind the app was simplicity, convenience, and efficiency. But I slowly learned that there are levels to convenience and if different products provide users with different levels of convenience, humans will always spend money on what makes their lives the most convenient. And that single thing is what made my app impossible to monetize.

The Importance of Market Feasibility When Building Anything

There were 2 main top dogs in the health and diet software niche: MyFitnessPal and Cal.ai. These 2 products are phenomenal and have attracted users and dominated the niche for a long time. They offer very similar and in some cases the exact same features that Calquest does. They track calories (Cal.ai), they track your water intake (MyFitnessPal), and they help you intermittent fast (MyFitnessPal, very recently). So essentially there were 2 top dogs, and many others, in this market niche that were killing it and killing it well. So the obvious question is “why even bother creating something to compete?” That’s the one question I should have asked myself before building anything. I did not do any market feasibility study to see if what I was building was even needed. I jumped head first into code and started building.

The Problem:

After 3 years (on and off) building this app, when it was finally time to release it and add a pricing model, I noticed something really interesting. My app uses Posthog for analytics and event logging so I had a clear view of my beta users’ interactions. I could see what features they used the most, how many seconds or minutes they spent on a specific page, and the most important metric of all, the DAU (Daily active users). After the first 2 weeks of the beta release, I noticed my Daily Active Users tanked. People weren’t returning to the app. People who returned were spending a lot less time on it. My notification interactions tanked. People weren’t even clicking notifications anymore. And that’s when it hit me that “I might have just built something nobody truly had the need for.” Before you say “Levi, maybe you picked the wrong beta testers,” the first 50 beta testers I hand picked myself. These are people I know are big into fitness and dieting and that was my target demographic. The issue wasn’t the demographic of the beta testers or users. It was bluntly that my app wasn’t providing any value to them. And it was all because of one feature. One freaking feature!!!

The Missing Feature:

Remember what I said earlier about convenience and how people will always spend money on what makes their lives most convenient?

But I slowly learned that there are levels to convenience and if different products provide users with different levels of convenience, humans will always spend money on what makes their lives the most convenient. And that single thing is what made my app impossible to monetize.

Well my app was missing one really important feature that would have made it so convenient and valuable to users that it would make sense for them to pay $19 a month for it. This feature is what MyFitnessPal, Cal.ai, and every other top dog in this niche did phenomenally well. And that feature is AI food recognition. Let’s say you had a bowl of chicken garlic pasta with a side of steak for lunch. Apps like MyFitnessPal and Cal.ai let you take a picture of your meal and they use an AI model trained to detect a vast category of foods to identify the meal, the portion, and then calculate how many calories it is along with the macros. Literal magic. This single feature made their apps so convenient that it was easy for people to pay $19 a month for it. It was a no brainer.

And there was me, trying to charge the same $19 a month for an app that made their lives less convenient. Even if I slashed the subscription in half to $8 a month, it still wouldn’t make sense. It’s like paying $8 for a Honda Civic knowing you can get a Tesla for $19 😂. A lot of you might say “why don’t you implement the same feature and add a similar price tag to the app?” Well, that feature is not as easy to incorporate as the other features.

Why Adding AI Food Recognition Was Impossible for Calquest

From a software engineering perspective, I like to categorize AI into 2 basic categories: Generation and Recognition. Of course there is reasoning, but that happens in both recognition and generation. For example, everyone has used ChatGPT to query something like “how many calories are in a McDonald’s burger?” and it returns an answer. That’s reasoning and generation. You’ve probably also sent ChatGPT an image or document to ask it a question based on that image or document, like “how many phones are in this image?” That’s recognition and reasoning. What’s my point? It’s significantly cheaper to perform reasoning and text generation on a model than it is to perform recognition and reasoning. Why? I honestly don’t know. I’m not an AI or ML engineer, but my guess is, compute time is different?.

I say all this to say that as a software developer, implementing AI text generation and reasoning like a chatbot is easier and cheaper, but implementing an AI recognition feature like what MyFitnessPal and Cal.ai have is significantly more expensive. How expensive? There are tools out there like Passio and CalorieMama that have AI recognition APIs using models specifically trained on food data. These tools cost a minimum of $2000 a month. Now if we do the math. Let’s say our app has an $18 subscription plan. We would need roughly at least 110 paid users a month to sustain the bill from these AI recognition APIs. Now 110 paid users doesn’t look like a difficult number to get to, but this is the issue. Before we can get even 1 paid user we still need to dump $2,000 into this API first, then start marketing the app and hope people actually find the app more useful than whatever they already use. Ideally it would take 2 to 3 months to get 110 paid users for a calorie counting app that just launched, considering the market competition. Also consider the app is still in beta and isn’t available to the public, so that timeline will look more like 3 to 4 months because you still need time to build on the API after purchasing it. So all in all, that would put me at a loss of close to $8,000 to $10,000 before getting my first paid user.

Being a broke college student, and I’m not even kidding, I’m actually broke 😪, it would be stupid and risky to dump $10,000 into one single feature for an app whose value is still uncertain given the current market dominance.

Lesson Learned

Market feasibility study!!! Market feasibility study!!! Market feasibility study!!! If your goal is to build a product and monetize it, it’s very important to understand the market and your competitors. This helps you build a product people will actually pay for and not even mind switching from your competitors to you. You’ll be way happier and better motivated when building something you’re confident has a chance to compete.

I still truly believe in my idea and I will definitely pursue it when it’s financially possible. There are so many features and integrations that are yet to be developed that I’m confident will increase the value Calquest has to offer to users.

Conclusion:

I just realized I did not even include AWS cost to the $2000 😭.