The AI tools stack that helped me go from $1k to $5k/month online (and what I'd cut)
Two subscriptions did most of the work, three more paid for themselves fast, and three others were dead weight I should have cancelled sooner.
AIFive months ago, this newsletter cleared about $1,200 a month. Last month it cleared just under $5,000. Nothing dramatic happened to the underlying business. The audience didn’t triple overnight, and there was no secret algorithm hack. What changed was the toolkit I used to run it, and just as important, which tools I quietly cancelled along the way. This isn’t a roundup of forty AI apps you should try this week. It’s the short list that actually moved revenue, what each one costs right now, and the three subscriptions I dropped because they were dead weight.
The two tools that actually carried the weight
Here’s the unglamorous truth: almost the entire jump came down to two subscriptions. Claude Pro for drafting and editing, and Perplexity Pro for research. That’s $40 a month total, and it’s still the core of everything I ship. 🧠 You can check current tiers yourself on Anthropic’s pricing page before committing to anything.
I tried ChatGPT Plus first, because everyone does. I kept it for about six weeks before cancelling. It’s a perfectly fine tool, and the major assistants have mostly converged on the same $20 price point anyway, so cost wasn’t what decided it. What decided it was fit: Claude handles long-form drafts and editing passes without losing the thread of a 2,000-word piece, and I stopped having to fight it to sound like a person instead of a press release. 📝
Perplexity earns its $20 by doing the boring part I used to burn two hours a week on: finding current numbers, checking whether a claim is still true, and surfacing where the actual disagreement is on a topic. It’s not a writer. It’s a fast, honest research assistant, and that distinction matters more than people give it credit for. 🔍
What this pair actually replaced:
Two to three hours a week I used to spend cross-checking facts by hand
A part-time researcher I’d been quietly considering hiring
The habit of publishing first drafts because editing felt like too much extra work
Roughly $80 a month I was previously spending on a grammar tool that did less
If you’re starting from zero, skip everything else and run just these two for thirty days before touching anything more. Depth before breadth beats a shelf full of half-used logins every time. 🚀
What got added once there was real money coming in
The stack didn’t stay at $40. Once actual revenue started showing up, three more tools earned a spot, and each one paid for itself inside the first month. 💰
Beehiiv runs the newsletter itself. The free Launch plan covers you up to 2,500 subscribers, so there’s no reason to pay a cent before you have an audience worth monetizing; full tiers are on Beehiiv’s pricing page. I crossed that line around month three and moved to Scale, which runs roughly $43 to $49 a month depending on billing. That upgrade unlocked the ad network and paid subscriptions, and the sponsorship math on a list this size isn’t complicated: general newsletters typically pull $20 to $50 CPM, and creators with a genuinely engaged list of a few thousand subscribers are commonly clearing a few hundred dollars a month from ads alone before affiliate income even enters the picture.
Canva Pro (roughly $13 to $18 a month depending on billing, see Canva’s plans) and Midjourney’s Basic plan ($10 a month) cover the visual side. Neither is exciting, but a newsletter with a real header image and clean pull-quote graphics converts better on social, and that’s a fact I resisted for way too long because I don’t think of myself as a “design person.” 🎨
If you want the fuller breakdown on what to actually build once you cross that subscriber threshold, three AI income streams that actually work covers the monetization side better than I have room for here. This is exactly the kind of topic the BizWhat Membership digs into properly, with a dedicated ebook and real implementation details. 📚
What actually earns money at this stage:
Sponsorships, once your list crosses roughly 2,000 to 3,000 engaged subscribers
Affiliate links for tools you already use and can vouch for honestly
One paid digital product, not five half-finished ones
Recommendation swaps with newsletters in adjacent niches
Have you actually priced out what your list could charge for a sponsored slot, or are you still assuming you’re not big enough yet? Worth doing the math before you decide that for yourself. 📈
The automation layer that saved the actual hours
None of this matters if it still eats your whole week. The stack above produces better content faster, but the real time savings came from automating the plumbing between tools, not from any single AI model. ⚙️
The setup is unglamorous: a spreadsheet holds approved topics, a simple automation watches for new rows, sends the topic to Claude for a draft, and drops the result into Beehiiv as a post ready for editing. No manual copying, no tab-juggling. I wrote up the exact build, connector by connector, in how to build a set and forget content pipeline, and it’s still running with almost no changes since I set it up. 🔧
What used to take a full day now takes closer to two hours: draft review, one editing pass, and scheduling. The automation isn’t writing anything on its own. It’s removing the fifteen small manual steps that used to sit between idea and published, which is a less exciting story than “AI does everything,” but it’s the honest one. 🕐
Where the time actually goes now:
About 45 minutes reviewing and editing each AI-drafted piece
20 minutes on visuals and formatting
15 minutes scheduling and cross-posting
The rest goes to client work or next week’s research
One honest caveat worth naming: 2026 is full of pitches for fully autonomous “AI employees” that supposedly run an entire function without you. Most of them oversell. The setups that actually hold up are hybrid ones, where automation handles volume and a person still catches the exceptions, which is exactly what this pipeline does and nothing more. 🤖
What I’d cut if I were starting over
Here’s the honesty part, because a stack list without the cuts is just an ad. 🗑️
Three subscriptions got cancelled on the way to $5,000 a month, and I don’t miss any of them. ChatGPT Plus got dropped once Claude covered the same ground for my specific workflow; that’s not a universal verdict, just what happened when I compared six weeks of output side by side. A keyword research tool I paid $99 a month for got replaced almost entirely by asking Perplexity sharper questions, which felt too simple to count as a real substitute, but the traffic numbers didn’t care that it felt simple. And I cancelled a social scheduling app that promised to handle everything and mostly generated posts that sounded like nothing I’d actually say.
The pattern across all three: I bought them before I had a specific bottleneck they were solving, based on someone else’s list of must-have tools. That’s the trap. The tools that stuck were the ones I added because a real task was taking too long, not because a thread online told me I was falling behind. 🎯
Tools worth skipping until you have a specific problem for them to solve:
Anything marketed as a fully autonomous “AI employee” replacing a whole role
A second or third AI writing assistant once one already covers your workflow
Premium image or video tiers before you’ve confirmed the visual actually helps conversion
Automation platforms bought before you’ve mapped the manual steps you’re removing
So if you’re weighing whether to add another tool this month, ask yourself one thing: what specific task is currently slow enough to be worth $20 or more to fix? If you can’t name it, that’s your answer. 🧭