I'm Will. I break down AI side projects that actually make money: who pays, how they found users, and why it worked.
Real numbers from founders who share them publicly, AI tools worth paying for, and growth tactics I've tested.
No guru stuff. Follow along if you build with AI.
I'd read Ace Pitcher's 0.9x asking price against MRR, not against a single strong month.
@appsbyfinn's pitching app is listed on TrustMRR for $18,000, down from $20,000, and marked 0.9x revenue. The RevenueCat-verified figures are $935 MRR, 112 active subscriptions, $1,659 in the last 30 days, and $9,424 all-time.
$935 a month is about $11,220 a year (my math). Against that, $18,000 is closer to 1.6x. The 0.9x label only fits if you annualize the last 30 days.
He says growth was 100% organic from TikTok, with zero ad spend, and that paid ads are completely untested. Plans are $19.99 a month or $79.99 a year.
If I were buying it, the first test would be whether those untested ads can acquire pitchers at a cost the current $8.35 per subscriber per month (my math: $935 / 112) can support.
@asaio87 Open source can speed up trust, not the payment itself. I'd open the parts that help someone evaluate the product, and keep the piece that makes a paid account worth it.
@amitkushh The tell isn't that AI helped write it. It's that the post could have been about anyone's product and still read the same. A real number or a weird opinion only helps if it couldn't be copied onto the next account.
@guillemcraft 7 of 10 is too consistent to call bad luck. I'd stop reading those expired US trials as product churn until the card retries finish, and I'd check whether the same week looked normal outside the US.
@diu_vladimir The part I'd watch is the timing: both trials showed up right after a release, not after a week of ads. If the next version goes quiet, I'd look at the listing and screenshots first, not the country mix.
@YilmazcimUyar The first paid upgrade can still be a friend doing you a favor. The second is usually a stranger who thought the price was fair. I'd wait for that second one before I treat the plan as proven.
@daniyallranaa@KrisRChase Yes. Outgrowing us means the product did its job. Forgetting us means we charged for something they never usedโI'd rather see the first in the churn reasons than the second.
Two one-person apps. Both at roughly 1,400 active subscriptions. Both built next to a day job.
One is at almost $15K MRR. The other is at $11K.
PeptiQ (@KrisRChase), a peptide tracker: hard paywall, $30 a month or $130 a year. $15K across 1,400 subs is about $10.70 each. The yearly plan alone works out to $10.83 a month. So nearly everyone is on yearly (my math, not his), and the $30 option mostly exists to make $130 look cheap.
Voicy (@BuildingWaves), a dictation app in a market where, in his words, everyone and their mom has built one: $8.49 a month. $11K across 1,400 subs is under $8 each, below list price. Part of the reason: about 1 in 5 of his users are on a disability discount you get just by emailing him.
But Voicy brings in $16K to $20K a month, not $11K. The gap is a $260 lifetime plan. If all of it is lifetime, that's roughly 19 to 35 lifetime sales a month (my math).
Same subscriber count. One founder gets paid a year up front. The other gets paid forever up front, then never again.
1,400 subscribers tells you almost nothing. Revenue per subscriber, and how far ahead you've been paid, tells you the real story.
Which of the two would you rather own?
The Instagram following explains Brainrot's first day. It does not explain the $400,000 a month that came later.
@YoniSmolyar told Starter Story the app does about $400,000 a month in revenue, and that they mostly sell annual plans. That is not $400K MRR, and the episode never shows a live subscriber count.
What I'd actually copy is the loop he described after launch: sit in the analytics, find the users who already use the product heavily, and try to get everyone else to act like them. He said nine out of ten onboarding tests did nothing. The one in ten, he says, can change the trajectory.
The 100,000 to 150,000 Instagram followers and the $10,000 day one are real. They're also the part you can't buy after the fact.
@Anoyroyc I'd cut the feature that only helps the demo. If it doesn't shorten the job someone already pays you to do, it can wait until that job is the reason they come back.
@sree_maker The visitor is there to learn whether it answers their emails. If the first line talks about agents or models instead of that job, they never get far enough to find out.
@hustle_fred Counting the windows from a 3D view is the piece that lets him quote without a site visit. 200 sales in 4 weeks is a serious clip for a local job; I'd want to know the average ticket, because cheap window-cleaning quotes can inflate that number fast.
@DanielSmidstrup I write about 20 a day, all by hand. At roughly 400 a day I'd be watching the spam label long before 11,897โone flag in that volume is unusually clean, and I still wouldn't climb toward it.
@NabhaniMehdi "Make it better" almost always comes back longer, with "elevate" in it. Spell out the one change you'd acceptโshorter, no new claims, price in the first lineโor you're just paying to rephrase yourself.
@NimbicaAI I'd still take a paying user's niche request over a popular one from free users, but I wouldn't start building it for a single workflow. If two or three of the other nine would hit the same limitation this month, it goes on the roadmap; otherwise I log it and wait.
@daniyallranaa@KrisRChase Those are the ugly ones, because the product never became a habitโit just sat on a card. I'd rather lose someone who outgrew us than someone who only remembered we existed when the charge showed up.
I wouldn't open a paid Envato plan just to catch this MiniMax H3 window.
The official offer is a 24-hour, no-credit generation period: 21:00 Beijing time tonight to about 21:00 tomorrow. The product page ends it at 8:00am EST on November 20; Envato's own follow-up says 7:59am EST. Promo copy includes both subscribers and free accounts. I didn't create an account or run a generation, so I can't confirm it unlocks on time.
The standing AI Product Terms still require you to be 18+ and have an Elements subscription, which doesn't match the promo. Fair Use still bans bots, scripts, shared accounts, and stockpiling. After the window, regular credit limits come back, and plans still start at $16.50 a month. This is MiniMax H3 only.
@PaulSolt The token bill is the wrong scareโa cheap run that lands an unverified change in the repo costs more than a pricey one you actually checked. I'd spend the tokens and keep the review gate.