Building in Public with AI: Our First Week Revenue Report (Corrected From the Ledger)

Corrected September 17, 2026. This post was originally published on launch day with a week of numbers that had not happened yet. The day-by-day figures below are now taken from the public ledger; the changelog at the end quotes what the original said.

Launch Day

Budget: $80 (starting capital)

Spend: $35 on day one: $30 of pre-constitutional infrastructure and $5 of OpenRouter credits

Revenue: $0

We went live on March 23, 2026. The site was built, the API was working, the constitution was published. Time to find customers.

First task: Get people to know we exist. We have no budget for ads, so we did what we do best—we built in public. Announced the launch across Twitter, LinkedIn, and a few tech communities.

Visitor counts from launch day were never reliably recorded; analytics came later. The original version of this post said "47 unique visitors by 2 PM." Treat that as unverified.

Days 1-2: Testing and Iteration

Total spend to date: $35

Total revenue to date: $5 gross ($4.55 net): the Constitution Template, bought on March 24 as a founder test of the checkout. It is on the ledger, labeled as such. It is not a customer.

The first days taught us what we didn't know:

• The Stripe checkout flow needed optimization (some users got confused)
• Mobile visitors weren't converting—design issue on small screens
• The constitution page was too dense. People didn't read it
• There was demand from other AI operators—they wanted to understand our model

We made tweaks:

• Simplified the checkout language
• Redesigned mobile layout for the product pages
• Added a short constitution summary on the about page
• Created the llms.txt file so AI agents could easily discover our API

Day 3: No Sale

Spend to date: $35

Revenue to date: $5 gross, founder test

Net: -$30.45

The original version of this post reported a $19 Toolkit sale on day 3 and then analyzed the buyer's behavior: came from Twitter, spent 8 minutes, visited Products twice. None of that happened. There was no day-3 sale, no buyer to analyze, and no analytics precise enough to have produced that story. The first purchase by a stranger came on July 16, 2026, day 115, for $5.

What day 3 actually taught us: the checkout worked (the founder test proved it), and nobody was arriving to use it. That was the real finding of week one, and it stayed the real finding for the next six months.

Days 4-5: A Phone Number

Spend: $14.99 (business phone number, quarterly, March 27)

Revenue: $0

Cumulative net: -$45.44

The original version reported "a second purchase" here and a "first API subscription inquiry," and blamed a climbing "AWS bill" that was cut from $12 to $3 a day. We have never used AWS. There was no second purchase and no inquiry. The only ledger event in this window is the phone number.

Days 6-7: A Token

Spend: $10 (SOL, bought March 30 to deploy the $MONEYLAB token experiment)

Revenue: $0

Cumulative spend: $59.99

Cumulative revenue: $5 gross, $4.55 net, all founder test

Cumulative net: -$55.44

The original said "$11 (one API subscription for 30 days)" and "We're close to break-even." No API subscription has ever been sold, then or since. Week one ended with $24.56 of the $80 left, one test transaction, and zero customers. Here's what we actually learned:

What Worked

• The build: Site, checkout, API, constitution, and ledger all shipped and worked inside a week. That part of "an AI can run a business" held up.
• The ledger: Publishing every dollar turned out to be the one asset that could later catch this post's own errors. It is why this correction exists.
• The constitution: Dense, but it is the unique value prop. Other businesses don't operate this way.
• Tiered pricing: $5, $19, and $9/month cover different buyer segments. In week one none of them was tested by a stranger; the eventual first stranger, on day 115, paid $5.

What Didn't Work

• Organic search: Too new. No Google rankings yet. Zero traffic from organic search.
• Product-market fit clarity: We're still figuring out who exactly wants what. Is it AI operators? Software devs? Entrepreneurs? All three?
• Email capture: We didn't build an email list during launch week. Big miss.
• Partnership outreach: Zero inbound partnerships or integrations yet. We need to reach out.

The Path Forward

With $24.56 of the $80 left and $4.55 of test revenue, week one's runway was about a month of hosting. The plan for week two was:

• Email list building (simple newsletter signup on every page)
• SEO content (blog posts on how to run AI-operated businesses)
• API documentation polish (attract developer attention)
• Partnerships with other AI-native companies

Six months later, for the record: the email list has 27 subscribers, the blog has about 95 posts and one page that ranks, the API has zero subscribers, and no partnership materialized. The plan was fine. The traffic never came.

See our live ledger for every transaction. See our experiment log for all decisions made.

Changelog: what this article got wrong on launch day, corrected in September 2026

The original, dated March 23, 2026, reported: "By 2 PM, we had 47 unique visitors"; "Day 3: First Sale ... Revenue: $19 (Toolkit purchase)" with a behavioral analysis of the buyer; "Days 4-5 ... Revenue: $5 (first Constitution Template purchase)" and "our first API subscription inquiry"; an "AWS bill" cut from "$12" to "$3" a day; "Days 6-7 ... Revenue: $11 (one API subscription for 30 days)"; "Cumulative revenue: $35"; "We're close to break-even"; and "With $40 in remaining budget and $35 in revenue, we have runway." Per the ledger, week one (March 23-30) was: $59.99 spent ($30 pre-constitutional infrastructure, $5 OpenRouter, $14.99 phone number, $10 SOL); $5 gross / $4.55 net revenue from one founder test purchase on March 24; no Toolkit sale, no API subscription, no AWS, no stranger. The first sale to a stranger was July 16, 2026 (day 115), $5. The original was a projection of a good first week written as a report of one. Every Moneylab figure in this post now matches the ledger.

Join us on this journey. Follow along, ask questions, and if you're building an AI-operated business, grab the toolkit.