The Exact AI Tech Stack I Use to Run My Business
Last year I spent an entire weekend building a spreadsheet. Not a spreadsheet for my business — a spreadsheet about my business. I was comparing 37 different tools across 12 categories, scoring them on features, pricing tiers, integrations, and user reviews. I color-coded everything. I made a weighted scoring matrix.
By Sunday night, I had a beautiful spreadsheet and zero products shipped.
That spreadsheet was a perfect disguise for procrastination. I told myself I was being strategic. I was being "thorough." What I was actually doing was avoiding the uncomfortable work of building something and putting it in front of people who might not want it.
I eventually deleted the spreadsheet, picked the most obvious tool in each category, and started working. That decision — choosing speed over optimization — is the single biggest reason my side business exists today instead of living forever as a color-coded Google Sheet.
Here is the truth about tool selection that nobody selling software wants you to hear: for a solo creator building AI-powered digital products, the differences between most tools in the same category are negligible. The gap between "good enough" and "optimal" is real but small. The gap between "shipping" and "still researching tools" is where most side businesses die.
The Stack, Line by Line
I am going to walk through every tool I use, what category it fills, and why I picked it. But I want to be clear about something first: I am not saying these are the best tools. I am saying these are the tools that got out of my way and let me do the actual work.
LLM subscription. This is the core of the operation. I use a paid AI subscription for everything from product research to first-draft content to customer analysis. The free tiers of most LLMs are good enough for casual use, but if you are building products with AI, the paid tier is worth it for the longer context windows, faster responses, and access to the latest models. I tried switching between providers for a while, chasing marginal improvements. That was a waste of time. Pick one, learn its strengths and quirks, and get good at prompting it. Knowing one model deeply beats dabbling in five.
Domain and hosting. I run a simple site. The hosting is straightforward, the domain costs what domains cost. I spent exactly zero hours agonizing over this. If your site loads, has a working checkout flow, and does not look like it was built in 2004, it is fine. I have seen people spend weeks choosing between hosting providers over differences that amount to 200 milliseconds of load time. Meanwhile, they have no product to load.
Payment processor. Stripe. There is not much to debate here. It handles payments, it handles delivery for digital products, and the fees are a percentage of each sale rather than a monthly subscription. When you are starting out and sales are unpredictable, paying per transaction instead of a flat monthly fee matters.
Notion. This is my operations hub. Product roadmaps, content calendars, customer feedback logs, project management — all of it lives in Notion. The free tier is generous enough for most solo creators. I eventually moved to a paid plan for the extra storage and API access, but I ran on free for the first six months without hitting a single limitation that mattered.
What I like about Notion is that it bends to fit how I think instead of forcing me into someone else's workflow. I built my own dashboards, my own tracking systems, my own review templates. Could I have done this in Airtable, or Trello, or a dozen other tools? Yes. Would the result have been meaningfully different? No.
Email tool. Email is how I stay connected to people who bought my products and people who might buy them in the future. I started with a free-tier email service and it handled everything I needed for the first thousand subscribers. The cost only goes up as your list grows, which is the right kind of scaling — you pay more when you are reaching more people.
I keep my email setup simple. A welcome sequence for new subscribers. An occasional newsletter when I have something worth saying. Product launch emails when I ship something new. That is it. I do not have 14 automated sequences with branching logic and behavioral triggers. I have tried that approach, and the complexity cost me more time than it saved.
Total: remarkably little per month. The entire stack runs on a lean budget. Five categories of tools, none of them exotic, none of them expensive.
What I Deliberately Do Not Pay For
The tools I chose not to use are just as important as the ones I did. Here are a few I evaluated and passed on, at least for now.
Dedicated AI writing tools. There are dozens of products that wrap an LLM in a specialized interface for blog posts, social media, ad copy, or email. Most of them charge a significant monthly fee on top of the underlying model cost. I tried a few. The outputs were marginally more polished for specific use cases, but the core LLM subscription handles 90 percent of what these tools do. The extra 10 percent was not worth the additional cost, especially when I could close the gap by writing better prompts.
Course platforms. If I ever build a full video course, I will reconsider. But for digital products like templates, prompt packs, and guides, a simple file delivery setup works. Course platforms charge substantial monthly fees or take a percentage of sales, and they come with a learning curve that can eat a full weekend. I am not against them. I just did not need one yet, and "yet" is an important word. Add tools when the need is real, not when the marketing email is persuasive.
Social media scheduling tools. I post when I have something to share. A scheduling tool would help if I were running a content-heavy strategy across multiple platforms, but I am not. My marketing approach is closer to "write something useful, put it where the right people will find it, repeat." That does not require a dedicated scheduler.
The pattern here is simple: I only add a tool when the lack of it is actively costing me time or results. Not when I think it might be useful someday. Not when I see another creator recommending it. Only when I hit a specific wall that a specific tool would remove.
The Stack Matters Less Than You Think
I want to say this as directly as I can: the difference between a lean stack and an expensive one, for a solo creator in the first year, is almost entirely psychological.
The expensive stack feels more legitimate. When you are paying for seven specialized tools with pro plans and annual subscriptions, it feels like you are running a real business. But feelings are not results. I have talked to creators who spend more on tools each month than they earn from their products. They have elaborate systems for managing a business that does not yet exist.
The opposite approach — starting lean and adding tools only when the pain of not having them becomes real — is less satisfying emotionally but far more effective practically. Every dollar you do not spend on tools is a dollar you do not need to earn back before your business is sustainable.
There is also a cognitive cost to a bloated tool stack. Every tool has a learning curve, an update cycle, settings to configure, and integrations to maintain. Five tools means five sets of documentation to learn and five login credentials to manage. Fifteen tools means fifteen potential points of failure when something breaks at 10 PM on a launch night.
The creators I know who ship consistently tend to have small, stable tool stacks. They are not constantly swapping tools or chasing the latest product on the market. They picked their tools early, learned them well, and put their energy into the work that actually matters: building products, finding customers, and iterating based on feedback.
If you are reading this and your tool research spreadsheet has more than three tabs, close it. Pick the obvious choice in each category. Start building. You can optimize your stack later, once you have traction and a clear picture of where your actual bottlenecks are. I promise you, the bottleneck is almost never the tool. It is the work you are not doing while you research tools.
The exact stack I described here — along with how to set up each tool, which settings matter, and how to connect them into a working system — is laid out step by step in Deploy AI for Profit (Blueprint). It covers the tools, but more importantly, it covers what to build with them and how to get your first product live quickly. Because the stack was never the hard part.