Last month I sat down to cancel one AI subscription and ended up finding six I had completely forgotten I was paying for. Two of them I had not opened in over ninety days. That is the moment I actually did the math on what running an AI-powered content and design business costs every month, and the number embarrassed me. If you are a solo creator or a small studio stacking AI subscriptions the same way I was, this is the honest audit I wish someone had handed me a year ago.
The stack I was actually paying for
Here is what showed up when I pulled my card statements for June: ChatGPT Plus at $20, Midjourney at $30 for the standard plan, ElevenLabs at $22 for voice cloning, Runway at $35 for the standard video tier, Canva Pro at $13, a transcription tool at $15, and a grammar and tone checker at $12. That is $147 a month, or just under $1,800 a year, before I even count the video editing software and cloud storage I was already paying for separately. None of these felt expensive in isolation. Stacked together they were quietly the second biggest line item in my business after rent on my workspace.
The uncomfortable part was not the total. It was that I could name maybe three tools in that list I used every single week. The rest were subscriptions I had signed up for during a specific project, told myself I would “probably need again,” and then let auto-renew on autopilot for months.
The test I now run before keeping any AI tool
I built a simple filter, and I apply it every 30 days without exception:
- Did I open this tool in the last two weeks, not the last quarter?
- Does it produce something a free or already-paid-for tool cannot?
- Would losing it change what I deliver to a client, or just how I feel about my workflow?
- Can I name the last output I shipped using it?
If a tool fails two or more of those, it gets cancelled that day, not “next month after I check something.” That single rule cut my stack from seven tools to four and saved me $64 a month, which is $768 a year I was setting on fire for the feeling of being well-equipped.
The three that survived, and why
ChatGPT Plus stayed because I use it daily for client emails, outline structuring, and pressure-testing scripts before I shoot. Midjourney stayed because reference imagery and mood boards for brand clients are a real deliverable I bill for, and nothing free gets close to the control it gives me over style. ElevenLabs stayed because voiceover work for UGC ad scripts is something I now sell as part of my product storytelling packages, and paying for one clean voice model is cheaper than a freelance voice actor per project.
Everything else got cut. Runway was the hardest one to let go of because it is genuinely impressive technology. But I was using it maybe twice a month, and the free tier plus occasional pay-per-generation credits covered that use case for a fraction of $35 monthly.
Why “it might be useful later” is the trap
Every one of the subscriptions I cancelled, I had justified with some version of “I might need this for an upcoming project.” That reasoning sounds responsible. It is actually just decision avoidance dressed up as planning. AI tools are proliferating so fast that there will always be a new one worth trying next month anyway. Holding onto last month’s “might need it” tool means you are paying rent on a maybe while a sharper, often cheaper alternative launches two weeks later. I would rather resubscribe for one project-specific month when I actually have the project than pay twelve months of insurance against a hypothetical.
The math that changes how you price
Here is the part that actually matters for anyone running this as a business rather than a hobby. If your AI tool stack costs $150 a month and you are only billing two client projects a month, that is $75 of tooling cost baked silently into every project before you have paid yourself a cent. Most freelancers I know, myself included for longer than I want to admit, never build that number into their pricing. You are either eating that cost as unpaid overhead or quietly working for less than you think you are.
Once I saw the real monthly figure, I started treating my AI subscription cost the same way an agency treats software licensing: as a line item baked into project quotes, not a personal expense I absorb out of enthusiasm for new tools.
A simpler way to think about it going forward
I now budget a fixed ceiling, $80 a month, for AI tooling. Anything I want to add has to replace something already in that budget, not stack on top of it. That constraint has made me a sharper buyer. Instead of collecting tools because they are impressive in a demo, I only add one when it directly replaces a cost or a task I am already paying for in time or money elsewhere.
The takeaways
- Audit every AI subscription against actual last-two-weeks usage, not intention
- Cut anything you cannot name a recent shipped output from
- Set a fixed monthly ceiling and force new tools to replace something, not stack
- Bake your real tooling cost into client pricing instead of absorbing it as invisible overhead
- Resubscribing for one active month beats paying twelve months of “just in case”
Frequently asked questions
How much should a solo creator budget for AI tools each month?
There is no universal number, but I have found $80 to $120 a month covers a genuinely useful stack for content and design work without paying for redundancy. The right number is whatever amount forces you to choose your best three or four tools instead of collecting everything that looks interesting.
Should I cancel a tool I only use for one specific type of project?
Not necessarily. If it is tied to a recurring paid deliverable, like voiceover for client video work, keep it and bake the cost into your pricing for that service. Cancel the ones you keep “just in case” with no attached revenue.
Is it worth paying for multiple AI tools that do similar things?
Almost never. Overlap is the fastest way this stack balloons. Pick the one tool per category that produces the best output for your actual clients and cut the rest, even if the others are good.
Related reading
- Why I Ditched CapCut’s Auto-Captions and Built My Own With Whisper
- I Tested AI Video Generators for UGC Ads So You Don’t Have To
- The Honest Guide to AI for Small Business Owners
If you are building a product and need someone to turn what it actually does into video people understand and act on, see how we do that.
Written by Osato Umweni, a designer and tech creator based in Lagos. More about me.
