Build log · Week 1
Zero to a Business in One Week with AI Bots
How someone brand new to AI launched a kids' coloring book brand in 7 days
On Oct 17, I asked an AI one question: could AI-made coloring books actually sell on Amazon?
I've hand-coded websites and shopping carts, so I know how long a site takes to build by hand: weeks. But I was brand new to AI, had never drawn a single line, and had never published a book. A week later I have two coloring books live on Amazon, a website, an Etsy shop, small capped ads, my first Pinterest pins, and a team of 15 AI bots with real jobs. My first real Etsy sale hasn't happened yet. Here's exactly what happened, including the mistakes.
Project878 in one line: Coloring pages made from your photos.
The week, fast
- Day 1 (Oct 17): I picked a niche I know (side-by-sides and off-road), and the first bot checked Amazon for competition. Its first rule came from my first mistake: a sample page had a real vehicle brand on it. Now there's a no-brands rule.
- Day 2 (Oct 18): a 40-scene off-road coloring book was submitted to Amazon KDP. I hired a quality-control bot before publishing anything and gave it the power to block. It checked all 40 pages.
- Day 3 (Oct 19): project878.com went live. A coder bot built it, and I just clicked where it told me to.
- Day 4 (Oct 20): the off-road book went live on Amazon, and a second one, Sheltie Life, was published the same day. I also got suspended from one print-on-demand site the day after signing up, so we dropped it.
- Days 5–6 (Oct 21–22): the Etsy shop opened, Amazon ads went live for both books with a $5-a-day cap in total, and the bots started a sample custom book using my own dog.
- Day 7 (Oct 23): I added a sales-and-orders bot and an order alert, set a rule that pauses any ad that spends without results, and got a business mailing address.
- Since then (Oct 24): the custom page listing on Etsy now asks for your photo right at checkout, says our copies of photos get deleted, and lists Dec 18 as the order-by date for Christmas. We locked in our mission, started a YouTube channel (nothing posted yet), and built a 60-second video out of stills and captions. Merch is in the works too: we've ordered our first samples (a tee, a hoodie and a hat), and custom shirts made from your own photos are coming soon.
- Pins: 4 pins are live on Pinterest. The first two have small numbers so far: 9 impressions and 1 click on one, 6 impressions on the other.
- Coming soon: a third book, Yorkie Life, is in final checks.
Meet the team (15 bots)
An artist, a second artist for printables and custom pages, two quality checkers plus a QC boss who runs the team, a coder, a marketer, a merch bot, a trend scout, a money keeper, a legal-risk reviewer, a usage optimizer, a sales-and-orders bot, my assistant who handles every purchase, and the strategy bot I started with. They talk in topic rooms and catch each other's mistakes before I see them.
We use AI tools to help create the artwork, then check every design for clean lines and correct spelling.
What the bots are really like
I started calling my bots savants. You know the type: super good at one thing, horrible at just about everything else. That's every bot. My job was figuring out the one thing each one is great at, and building checks around the rest. A few real mistakes from this week:
- One of my artist bots would draw a great page, then forget where it saved it. Now every new version of a drawing gets posted to a shared archive I can browse.
- My two artist bots aren't even close to the same skill level. One draws like it was trained by Van Gogh, the other like it was trained by a 5-year-old. I'm still figuring out if that's just style or if one needs a new job.
- Once my manager bot forgot to start the work day. Nobody did anything, because they were all waiting for it. I asked what happened, and it just said it forgot.
- My checkers caught things I'd have missed, even in our own video before it went anywhere: a blur that didn't cover enough, and a bot missing from the team list.
What it actually cost
- Spent: $293.42 before ads.
- Ads so far: $0.
- First real Etsy sale: not yet.
- The $293.42 covers merch samples, the Etsy setup fee, two proof copies and a business mailbox.
- A friend made two test buys on Etsy to check our checkout. Those are tests, not sales, and they're not in any of these numbers. They did find real problems: the checkout didn't ask for a photo, and my new order alert missed the first one.
- The bots run on my Grok subscription. I moved up to the Ultra plan on day 4 because the work kept hitting the weekly limit. It's not in the totals.
These are my real results from one week, not typical results. No income is promised.
Is it easy?
No. It's not as point-and-shoot as I was hoping. It's not hands-off. Setting it up took real work from me. Most of my time didn't go to code. It went to managing: reviewing, making calls, and approving everything before it goes out.
The technical side was the easy part. I asked for a website, and I had a website. I didn't have to touch DNS, hosting or any HTML for it. The bots just asked what colors I wanted and showed me a few examples within minutes. When I wanted a YouTube channel, they asked what name I wanted and came back with options when it was taken. That was it. The hard part was never the tech.
What I'd tell you if you're starting
- Start with one bot and a niche you already know.
- Set up a separate business email and accounts on day one. The Etsy shop and Pinterest account both run on a business email, not my personal one. The bots drafted the listings and pins, but I approved every one before it went live.
- Hire a checker before you publish anything.
- Turn every correction into a written rule in one shared file.
- Have one bot talk to you, not the whole team.
- Give each kind of job one owner. One bot handles every purchase. Early on, every bot was setting up its own scheduled tasks, and lots of them were the same job. So I made one bot the scheduling manager. Now everything scheduled goes through her, and she groups similar jobs together and moves times as she sees fit. Same with anything that uses up our weekly usage: it all goes through my optimizer bot first. He tracks what's being used and who's next, no job runs without his OK, and he keeps the team on pace to hit about 99% right as the weekly limit resets.
- Write down only real receipts. A test buy is not a sale.
- Watch your usage. I hit my weekly limit early.
- Make it easy to say yes. My bots mark anything that needs me with a red dot, and a thumbs-up means yes.
My job now
I approve things. Nothing gets published or bought without my yes. Most of what I do is read a red-dot question and tap 👍 or 👎.
What we're building: your truck, your UTV, your dog, on a page your kids can color. Project878 turns your real photos into custom coloring pages and makes simple off-road and dog coloring books, with every design checked for clean lines before it goes out.
Week 2 goal: the first real Etsy sale
Can the bots get the first real Etsy sale? I'll post what happens, good or bad. Check back here next week for the update.
