2024 review, 2025 plan
My review of 2024 progress and my plans for 2025
It’s already February 2025, so I’m a bit late on this annual review, but I wanted to regroup from the past year and set some intentions for the new one.
In 2024, I set a few goals for my main project, Live Poker Theory:
Monetize the web app
Create a native version of the app and monetize that
Start a YouTube channel as a marketing lead
Improve SEO on the website
Publish a poker strategy book as a combination revenue source + marketing lead for the software
The good news is that I accomplished all of these goals.
The web app was monetized and got about a dozen paying users.
The native iOS and Android app were launched and got a few paying users
The YouTube channel was launched, and got 180 subscribers, and my most recent video was my best-performing one yet
I’m now the #1 app, and #2 overall result after some Reddit threads, if you Google search for “gto preflop trainer”, which gives me about 5 new free user signups every day
The strategy book was written and got a couple dozen sales
So overall, my process goals were all hit, but unfortunately, the outcome goals were not, as total revenue has been disappointing. The web app has churned paid users about as quickly as it acquires them, and the native app has only converted a handful of users so far. Overall, MRR is in the low hundreds, and conversion rates from free to paid users are abysmal. The YouTube channel is fun, and I think it has a lot of long-term potential as I improve at editing and understand what type of content works, but without the likelihood of a viral video, it’ll still be a long, slow grind up to a meaningful audience. Likewise, I had modest expectations for the book sales, but so far, I’ve underperformed those expectations.
My big-picture goal of turning Live Poker Theory into a sustainable business has failed, so far at least. However, there’s a foundation established to grow it further, and I still believe it has potential for significant revenue. I’ve only built a small fraction of my overall vision for the product, and I have countless ideas on how to improve conversion rates. But it’s challenging to evaluate how much effort to keep pouring into it when the ROI hasn’t materialized yet. There’s a high opportunity cost to all the other exciting projects I could be working on.
A lot of the effort involved with building Live Poker Theory can also be re-fitted to new projects. Having a scaffold that I wrote for monetized web apps and native apps means I can ship new apps with the same scaffold quickly. Likewise, writing a book, working on SEO, and making YouTube content are very valuable experiences to apply to new projects. But there’s still a seemingly infinite number of new ideas to build but very little clarity about which ones could meaningfully move the revenue in the least amount of time.
So, I am at a crossroads in terms of what work to prioritize in 2025.
The AI Elephant In The Room
There’s been breathless hype around LLMs and generative AI the last couple of years, but I’ve resisted doing AI to be part of the cool crowd because I didn’t want to work on anything that I wasn’t excited about for my own use cases.
At various points in my Computer Science undergrad and my time at Google, I had enthusiasm for topics around machine learning and its applications, but I found that motivation waned eventually unless I applied it to a project I was interested in. So, I didn’t want to work on any project that I wouldn’t stick with out of long-term interest.
I was also concerned that it’s already difficult enough to monetize apps, and if you extensively run AI models, the API or GPU costs can be significant - now you need a baseline of revenue just to break even, or you’re losing money on the operation. At least with Live Poker Theory, the revenue is effectively all profit besides the time invested.
Still, I’ve experienced FOMO over not working enough with this clearly revolutionary technology. It feels foolish to be doing software at a time when the entire industry is getting upheaved and not working with it in any way.
How Claude Helped Me Write a Better Book
My own interest in applied AI has organically grown as I’ve extensively used ChatGPT and Claude to help me develop my apps and write the book over the last year. When it came to writing the book, I didn’t use any AI generated text directly, but I did extensively use AI to help brainstorm the structure, the topics of each chapter, and the overall transitions.
For example, I had a Reddit post that was really well-received and one that wasn’t well-received. I fed both chapters into Claude, and Claude helped me identify that the well-received post started with a clear, surprising insight, then worked backwards through the solvers and the theory to justify it. The poorly received post just moved from topic to topic in a text-book style way. With that information, I went through each chapter in my book and made sure to put interesting insight up top and work backward from there in the rest of the chapter.
More recently, I’ve been seeing how tools like Cursor, Cline, and Windsurf can dramatically accelerate development speed - a topic I’m genuinely interested in as a solo dev whose biggest asset is time and would love to move faster.
There are a few other AI topics that interest me:
Using LLMs to help be a better creator. I recently learned you can give Claude a YouTube thumbnail you like, and it’ll help you re-create it in image software like Figma.
Using LLMs as a sort of life coach. One of my big passion projects is Navigoals - a habit tracker that makes it easy to track lots of data points about your data - and Interstitch , a time tracker optimized for personal usage. I mostly built these to keep track of my own life, but my recent passion project is feeding all this data into LLMs and getting coaching. I have a follow-up post planned on this topic.
So, while I didn’t want to work on AI for AI’s sake, my interest in LLMs applications has organically emerged.
Content is King, or, Ignoring SaaS Blinders
The other conclusion I’ve come to over the last two years is the power of content for online entrepreneurs. There are many people who are just doing content, notably known as influencers or creators.
Creators can make money from ad revenue on platforms like YouTube, through sponsors of their content, and through course sales. You also build an audience which is helpful for marketing, but perhaps less obvious,but just as important, that audience can serve as an inbound stream of conversations about what business ideas are even viable. This additional validation can reduce the risk of building something nobody wants.
Many people with software engineering backgrounds get what I call “SaaS blinders” - thinking that building a SaaS product is the only path to an online business . If you build software for a living or you want to start an online business, selling software seems like the obvious choice. But I’m increasingly seeing that while SaaS probably has the highest upside, it’s also one of the most difficult to get the initial momentum compared to content. Content may be a better “step-ladder” approach, where you build a small revenue source to make entrepreneurship sustainable before going after the bigger opportunities.
Putting It Together
Putting together my two ideas - 1) I want to explore AI without overcommitting to an unvalidated idea with high operational costs, and 2) I want to do more content; my conclusion is to work on YouTube content about using AI to make apps.
This approach will let me work on interesting apps without as much pressure to complete a fully fleshed-out product that converts to paid users, build an audience for more inbound ideas to work on, and potentially be a viable business of its own. I’m still leaving the door open to revisit SaaS if my content is mostly about building SaaS.
I also really enjoyed the YouTube work I did this year, particularly once I realized that YouTube will show new videos to random people, and, if it does well, show it to more people. There’s discoverability built in, and the cream rises to the crop more than on other platforms where you have to do “tricks” to bootstrap that early momentum.
For example, on Twitter if you create a brand new account and write a tweet, nobody will see it. People do things like reply to others to get noticed and grow their following before a regular tweet they write doesn’t go “into the void”. But on YouTube, if you start a brand new account and post a video, at least a few people will be shown it, and if they watch it, more people will see it. This makes it a more appealing platform to focus on for audience-building since you can just focus on content.
My biggest revelation in learning the above lesson was the following. I was cross-posting my YouTube videos to Reddit, and they’d usually get hundreds of upvotes on Reddit, which led to some YouTube views. But one post got downvoted on Reddit, so I deleted it. But that was my best-performing YouTube video yet! This is how I realized:
If you want to grow on YouTube, you just need to focus on making good YouTube content. If the watch metrics are good, YouTube will recommend you more. You don’t need to bootstrap your YouTube audience from another source.
The type of content that does well on one platform can be completely different from what does well on another platform. The deep strategy dives were loved on Reddit but not well received on YouTube, but the more “off the cuff” fun analysis was better received on Youtube despite being downvoted on Reddit.
So all of this is to say, I’m excited about:
Using LLMs to make web and mobile apps faster
Using LLMs and generative AI to make better content
Using LLMs as a personal life coach based on my own personal analytics I feed to it (more on this in the future)
Making video and written content about all these topics
Of course, this means I have plenty of interesting ideas to work on, but not a lot of clarity on prioritization.
For the next few months, my plan is to simply make YouTube content about these topics, especially on using LLMs to make full-stack apps faster. While it’s a competitive content category, it kills two birds with one stone for me in that I can use Live Poker Theory as the case study “app” to improve, while turning my day to day work into more technical content will hopefully open up new opportunities.
It’s also a low commitment strategy in case a more obviously compelling opportuntity emerges.
This Newsletter
I started this newsletter about a year ago because I always enjoyed writing and I was interested in the general trend of newsletters and Substack. But I was a little unsure what topic to write about.
I thought having a “build in public” newsletter would be something that I could consistently write , and was an often-recommended thing to do for “indie hackers” to share their story.
However, I’ve found that “build in public” is not a topic that I enjoy writing about on a monthly basis. I think the “build in public” advantages are a bit overstated - sharing updates on my poker app is interesting to other “indie hackers” but not my target audience of poker players. Not only does this not grow the target audience I care about, it might actually backfire if I just inspire more indie hackers to build poker apps and compete with me.
Another problem with the “month status update” posts is they often lacks a “give” to the reader if it just recants what happened over a month, rather than focusing on the post sharing a useful insight for the reader.
Importantly, I find it’s an unhelpful pressure to overshare every detail of projects I’m working on, while I’m still in a process of rapidly shifting priorities and experiments.
Going forward, I plan to re-organize my various pieces of writing.
This newsletter will still sometimes include updates on the business progress but will more generally be a “catch-all” sink for topics I’m interested in writing about without pressure always to share business updates. For example, I’m excited to write about the AI life coaching system I’ve been building, and my various experiments with using AI to help me make YouTube videos.
Overall, I had a lot of goals that weren’t as neatly achieved as I planned. My initial plan of this newsletter to neatly document a steady progress to $10k MRR was not achieved. But I’m as broadly excited about online entrepreneurship as ever. I just think my approach needs to be re-calibrated. And so, in 2025, I recalibrate
Summary
In 2024, I launched my first monetized web app, monetized native app, YouTube channel with decent viewership, and my first published book. While the foundation is promising, revenue is still low and the project is not yet viable.
In 2025, I want to spend more time exploring AI projects, and my avenue to do so will primarily be through creating content on building apps fast with AI. I’m especially interested in YouTube.
I’m also interested in using AI to be a more effective creator and as a life coach , particularly in a passion project where I collect a large amount of personal analytics that I can feed to an LLM
One of my biggest lessons in 2024 was that YouTube is one of the better places to try to grow as a new creator since content from unknown creators will get tested on viewers and recommended more if it’s watched.
Another big lesson was the content that does well on one platform might not do well on another - my best Reddit post were my worst YouTube videos, and vice-versa