Quick math up top, because that's the deal here.
If you've typed some version of "ai side hustle for stay at home moms" into a search bar hoping for a shortcut, I get it, I typed something close to that too. There isn't a shortcut, but there is a real number: $1,398 a month, from two small retainers, after tool costs, working about six hours a week around two kids under six.
Client one pays $800 a month for a content and SEO cleanup retainer, up from $550 when I started. Client two pays $650 a month for a similar setup I built after the first one worked. That's $1,450 in revenue. ChatGPT and Claude subscriptions run $40 a month combined, and a video tool I use for pitches is another $12. Net is $1,398, and that's the number I actually bank.
I'm Priya, 34, in Chattanooga. I've been home with my kids since my second one was born, and until this spring that felt like a settled decision. Then my husband's overtime got cut at his job, the kind of quiet schedule change that doesn't show up as a headline anywhere but absolutely shows up in a grocery budget. I didn't want to put the kids in daycare to go get a job. I wanted found money that fit inside the hours I already had, which were mostly two nap windows and whatever was left after bedtime.
The first three weeks were rough in a specific, countable way. I sent eleven proposals on Upwork for content and SEO work, jobs I'd never technically done before, and heard back from exactly zero of them. I nearly stopped there. What changed things wasn't a better resume, since I didn't have one for this. It was switching from a text proposal nobody reads to a two-minute recorded video where I just talked through how I'd approach their specific job. Proposal seventeen got a reply. That one turned into the first retainer.
One thing before I get into what actually worked. I spent about two weeks watching freelancing YouTube videos trying to figure out how to even write a decent Upwork proposal, and every video gave conflicting advice about rates, portfolios, and whether cold applying even works anymore in 2026. A short course cut through that fast, in an afternoon instead of two weeks of contradictory opinions. YouTube is fine once you already know what you're looking for. It is a maze when you do not.
One more thing, and it's the part that actually matters most for anyone doing this with small kids at home, so don't skip to the end without reading it. The money isn't the hard part. Fitting six focused hours into a week that's already full is, and there's a specific way I made that work that I'm saving for last.
I stopped applying with a resume I didn't have
I'd never run an SEO cleanup or built a content pipeline before this. What I had instead was five years managing a dental office's scheduling and patient communication, which is mostly problem-solving and juggling competing priorities under a deadline, just aimed at a different problem. So instead of pretending I had SEO experience I didn't have, I pitched the transferable part honestly: I told my first client upfront that I was newer to this specific type of project but not new to solving operational messes fast, and that I'd be using AI tools to close the technical gap quickly rather than learning it slowly on their dime.
That honesty, oddly, is what got the callback. Clients on Upwork have been burned by people who oversell and underdeliver. Someone who says "here's exactly what I don't know yet and here's my plan for it" reads as more trustworthy, not less.
The two-minute video that got replies
Eleven text proposals got silence. The one thing I changed for proposal twelve onward was recording a short video instead, just me talking through their specific job post for about two minutes, unedited, first take usually. It is a strange thing to get used to, hearing your own voice back, but it worked in a way text never did. My guess is that a client scrolling forty identical-sounding proposals actually stops on the one where a real person is talking to them about their exact problem instead of a generic template.
Where Claude actually did the work
Once I had the client, the honest part came due. I used Claude to build the actual SEO audit and content workflow I'd promised, working through it the way you'd work with someone more experienced than you, asking it to explain the reasoning, not just hand me an output I couldn't defend if the client asked a follow-up question. I checked everything against the client's actual site before sending anything. It was slower than I expected the first time and faster every time after.
Raising the price once the results were real
After about two months, the client's organic traffic had moved in a way we could both see in their own analytics, not something I was claiming, something they could check themselves. That's when I asked for $800 instead of $550. I was more nervous about that conversation than I probably needed to be. The worst case was they'd say no and I'd stay at $550, which was already more than zero.
I'd been using Claude in a scattered way before all this, ask a question, get an answer, move on to the next thing without much of a system. What actually helped was a short course on Mindwand that walked through prompting for real client deliverables in 15-minute lessons I could do during a nap window. Skool has communities covering similar ground, and Coursera has structured courses too if that format works better for you. Pick one. Finish it.
The part I saved for last: the nap-time math
Here's the actual schedule, because "a few hours a week" doesn't mean anything without the specifics. I work during two nap windows most days, roughly 45 minutes each if I'm lucky and closer to 20 on the days I'm not, plus one focused hour on Sunday nights for invoicing and planning the coming week. That's somewhere between five and seven hours most weeks, and I don't let it go higher than that on purpose.
The reason for the hard cap isn't discipline, honestly, it's self-preservation. The first month I let client work bleed into every spare five minutes, phone in hand during actual playtime, and I hated how that felt more than I liked the extra money. Now the phone goes on do not disturb outside the work blocks, and if a client needs something outside that window, it waits until the next one. Two clients have been completely fine with this. I'd guess a client who isn't fine with it isn't a client I want long term anyway.
If you want the structured version of the prompting side of this, Mindwand has a course on it, alongside the same Skool and Coursera options I mentioned earlier.
What $1,398 a month actually costs me
At six hours a week, that's something like $54 an hour on the client work itself, which sounds better than it feels some weeks, since it doesn't count the applying, the video recording, or the three weeks of silence that came before any of it paid anything. Averaged across the whole four months since I started, counting the dead weeks, I'd put the real number closer to $30 an hour. Still solid for something that fits inside naps.
I don't think this replaces a full income, and I wasn't trying to replace one. It's real grocery-budget money that showed up because I finally sent a video instead of another ignored text proposal, and because I decided in advance how many hours I was willing to give it. If you're weighing whether this fits your actual week, start with the hours question before the money question. The money follows once you know what you actually have to give it.