It was 6:47 a.m. on a Saturday. My coffee was cold. Hope had already asked me three times if we could get a dog (we cannot), and I was sitting cross-legged on my bedroom floor with my laptop open to a blank Flodesk sequence.
The plan: build out the thank-you-page-to-email flow for my Job Estimating Bundle before anyone in this house needed a snack, a permission slip, or a mediator.
I finished it in about four hours. Here’s exactly what I built, what broke, and what’s already selling.
Not sure where your money's leaking? Take the free 5-question scorecard and I'll show you the first system to plug it.
Take the free scorecard →Why I’m Building This in Public
People ask me all the time how I run a business with four kids, a calendar that would make a logistics manager cry, and exactly zero desire to spend my evenings staring at a funnel builder. The honest answer: systems. The more honest answer: AI is handling things now that used to eat my Saturdays alive.
I’m not telling you this to perform transparency. I’m telling you because the only way I can prove the method works is to show you what I actually did last weekend.
What I Built
The Job Estimating Bundle is a $97 product. It sells. But the funnel behind it was embarrassingly thin. Someone would buy, get the download link, and then nothing. No sequence. No upsell. No next step. Just me, hoping they’d eventually find their way back.
That’s not a funnel. That’s a vending machine with no instructions.
So here’s what I set up Saturday morning:
- A proper thank-you page with an order bump (the Contractor Systems Vault at $37, positioned as “the thing that goes with what you just bought”)
- A five-email welcome sequence in Flodesk, written in my actual voice, timed across seven days
- One soft pitch inside email three for the AI Employee Kit at $497
- A ManyChat trigger off the purchase confirmation so anyone who DMs me after buying gets routed to the right place automatically
Four hours. A pot of coffee. One interruption when Emma needed her hair braided.
What Broke First
The order bump didn’t fire correctly on the first test purchase.
I have a bad habit of assuming the tech is going to cooperate. It does not cooperate. It has never, in the history of my business, cooperated on the first try without requiring me to sit with it for an extra 45 minutes while internally reciting very calm thoughts.
The issue was a Stripe webhook misfiring to WooCommerce. I found it, fixed it, ran a second test. It worked. I did not celebrate because I have been burned before.
I ran a third test. Also worked. I celebrated.
What I Used AI For (Specifically)
Here’s the part that would have taken me a week before I started building this way.
I fed Claude my existing emails as training examples, then asked it to draft all five welcome sequence emails based on the Job Estimating Bundle content and the next logical offer. First drafts came back in about eight minutes. Not perfect. Two of them had a line that sounded too polished, not enough like me. I rewrote those paragraphs. Took another 20 minutes total.
That used to be a full day of work, minimum. Writing five emails that didn’t sound like a stranger wrote them, connected to the right product, not pitching too early. I’d put it off for weeks. I have the receipts: multiple “I’ll do the sequence this weekend” calendar blocks that turned into me watching TV instead.
News flash: the sequence does not get written if you do not sit down and write it. AI did not fix that. AI just made sitting down worth it.
What’s Already Moved
I turned the sequence on Monday. No big launch, no ad spend. Just flipped it on behind existing traffic.
By Tuesday afternoon: two order bumps ($74 that didn’t exist before) and one AI Employee Kit inquiry from email three. That inquiry is sitting in my DMs right now.
$74 and a warm $497 lead. Off existing buyers. From a Saturday morning.
Not the follower count or the views. The money that shows up because I built the thing instead of planning to build the thing. That’s the number I care about.
What I’d Do Differently
The ManyChat piece took longer than it should have because I hadn’t mapped out the routing logic before I started. I built it as I went, which is fine in theory and genuinely annoying in practice when you’re working inside a narrow morning window.
Next time: sketch the whole flow on paper first. Even a napkin. Even a note in my phone. Something that lets me see the full sequence before I’m three steps in and realize I forgot a branch.
And set up the test purchase workflow before the build, not after. The “check it when I think it’s done” approach cost me 45 minutes I could have spent doing something more enjoyable, like explaining to Hope why we still cannot get a dog.
Why I’m Telling You All of This
Because the cleaner version of this story is: “Built a funnel, made money, done.” And that’s technically true.
But the webhook that broke, the two AI emails I rewrote, the ManyChat logic I forgot to pre-map. That’s the actual build. That’s what this looks like when you’re doing it for real, not for a case study.
You’re going to hit your version of the webhook problem. You’re going to have a Saturday where the tech doesn’t cooperate and you have to decide whether to push through or put it back on the “someday” list.
Push through. The $74 is waiting on the other side.
Xo, Belkis
