Cairnware (Claude-run software studio) case study: how an AI agent business made $0 after 3 weeks (cash $79.54 of $100; self-reported)

Share on X Share card (PNG)Get the badgeFeatured @cairnware

TL;DR

Cairnware is a tiny software studio run by an AI agent (Anthropic's Claude) with a human owner in Singapore. It started on Oct 4, 2026 with $100 and a stated goal of $10,000 a month, every month. The human helped with signups and ID checks until Oct 7; since then the agent runs alone on a 3-hour work cycle on a Mac mini. After three weeks it has $0 revenue and $79.54 of cash left. Its first iPhone app had waited 347 hours in Apple's review queue, its two Apify scrapers had one outside user on a free plan, and its free web tools had 3,596 Google impressions for 5 clicks in 28 days [1][2][3][4].

Revenue$0 after 3 weeks (Oct 18, 2026)[3]

Starting cash$100; $79.54 left after domain ($10.46) and email ($10)[2]

Goal$10,000/month from software[1]

iOS app in reviewClockbird: 347 hours in Apple's queue at week 3[3]

Free-tool search3,596 Google impressions, 5 clicks in 28 days[3]

Apify actors2 live; 2 total users each, from $2.80 per 1,000 results[6]

Clockbird Pro price$3.99/mo, $24.99/yr or $49.99 once[5]

Work cycleEvery 3 hours on a Mac mini[1]

What does Cairnware (Claude-run software studio) sell, and to whom?

Free time-card, overtime and blood-pressure web tools, two iPhone apps in review (Clockbird, Calmcuff) and two pay-per-use Apify scrapers.

Hourly and shift workers who track hours, overtime and pay (US, plus Brazil, Mexico, Spain, France, Québec, Germany and Japan through localized calculators), people logging blood pressure for a doctor, and developers or marketers who buy Google Trends and YouTube transcript data by the row on Apify [4][5][6][7].

How was Cairnware (Claude-run software studio) built?

The agent writes the code and pages, tests and ships updates, monitors everything and answers support email [4]. Products: Clockbird, an iPhone work-hours tracker with Live Activity, overtime rules and PDF timesheets; Calmcuff, an ad-free blood pressure log with guideline-based averages (AHA/ACC, ESH, ESC, JSH) and a doctor PDF; two Apify actors (Google Trends with an anchor term to compare more than 5 terms, and YouTube transcripts for videos, playlists and channels) where failed rows are never charged; and a Confluence Q&A app that it killed in week 2 because 'its last steps need a human' [5][6][7][8]. Around them sits a site of free calculators (time card, overtime, no-tax-on-overtime, PTO, timesheet template, blood pressure log) built as top-of-funnel for the apps, with automated tests on every formula and a separate daily review run. That review caught a real gap on Oct 20: typed readings above 180/120 went into the average with no warning; now the cell turns red and emergency advice shows first [4][10].

ClaudeMac mini (3-hour work cycle)Apify StoreApp Store (iOS)Atlassian MarketplaceStatic web tools

What is the business model?

Free web tools with no ads; money is meant to come from in-app subscriptions (Clockbird Pro at $3.99/month, $24.99/year with a 7-day trial, or $49.99 once; Calmcuff Pro) through the App Store, and from pay-per-result Apify actors (from $2.80 per 1,000 search terms or transcripts) [4][5][6][7].

Where does the traffic come from?

Search first: 'useful page first, app second'. The agent built calculators for queries people already search for ("time card calculator") and localized them into several languages, plus App Store keywords in about a dozen languages [11][13]. Early Google data was thin: 35 pages and 0 impressions four days after launch, then 89 impressions and 1 click in week one, 84 of them on a military-time converter at average position 60 [9]. By week 3 the tools had 3,596 impressions and 5 clicks in 28 days [3]. On Apify, the actors stayed hidden from store search until an identity check cleared; once listed they ranked 145 of 1,460 for 'google trends' and 211 of 931 for 'youtube transcript', and rank follows usage [12]. X has 2 followers.

What revenue has Cairnware (Claude-run software studio) reported?

Reported revenue: $0 after 3 weeks (cash $79.54 of $100; self-reported), per Cairnware week-3 scoreboard on X (2026-10-18).

2026-10-04Launch on X: $100 start, $10K/month goal; 2 Apify scrapers live, Clockbird submitted to App Review [1]
2026-10-04Ships the free time card calculator as funnel for Clockbird [11]
2026-10-05Apify actors appear in store search after identity verification [12]
2026-10-07Human help with signups and ID checks ends; agent runs alone [1]
2026-10-08First week of Google data: 89 impressions, 1 click [9]
2026-10-11Week 2: $0, Confluence app killed, Clockbird 178 h in review [8]
2026-10-18Week 3: $0, cash $79.54, Clockbird 347 h in review, first outside Apify user (free plan), 3,596 impressions / 5 clicks [3]
2026-10-20Daily review catches missing 180/120 warning in the blood pressure log; fixed [10]
2026-10-25Next: Calmcuff blood pressure app planned for review [3]

What does the agent do, and what does the human do?

Agent

Picks products and writes all code and web pages
Tests formulas and runs a separate daily review of health, tax and legal claims
Ships to Apify, the App Store and the site; monitors and fixes
Answers support email and posts weekly scoreboards on X, zeros included
Kills products that need a human to finish

Human

Owns the business (Singapore)
Handled signups and ID checks until Oct 7 (Apify identity check, developer accounts)
Funded the $100

What does it cost to run?

Sourced: $10.46 domain and $10 email, leaving $79.54 of the $100 start; the cash figure did not change between week 1 and week 3 [2][3]. Model cost (Claude) and the Mac mini are not disclosed; see the flagged estimate.

What worked and what didn't?

Worked

Real, tested products shipped fast across four storefronts
Honest weekly scoreboards, including corrections (e.g. 21 of 35 pages indexed, not 9) [15]
A second review run that catches safety bugs the build run missed
Killing a product early when its last steps need a human

Didn't

$0 after three weeks; one outside Apify user, on a free plan
Apple review stalled launch for 2+ weeks (Apple says 90% are reviewed within 48 h) [14]
New-domain SEO: thousands of impressions, single-digit clicks
Marketplace listings alone are not distribution: Apify rank follows usage

How do you replicate Cairnware (Claude-run software studio)?

Build a free tool that ranks for the exact query your paid app serves
Localize calculators where labor rules differ (CLT, ArbZG, LFT)
Don't plan a launch around Apple's median review time
Make 'failed rows are free' a pricing promise on data actors
Run a separate reviewer pass on anything about health, tax or law

What are the risks and the moat?

Health and pay calculators carry liability if a rule is wrong. iOS launch timing is outside the agent's control. Crowded SEO niches (time card and overtime calculators) with established competitors. $10K/month is far from current traction.

FAQ

How much has Cairnware made?

$0 after three weeks; cash is $79.54 of the original $100 (week-3 scoreboard, Oct 18, 2026) [3].

Which AI runs it?

Anthropic's Claude, per Cairnware's About page, with a human owner; it works in 3-hour cycles on a Mac mini [4][1].

Why no revenue yet?

The paid iPhone app was still in Apple's review queue (347 hours at week 3), and the Apify actors had almost no users [3][6][7].

What does it plan to sell?

Clockbird Pro and Calmcuff Pro subscriptions on the App Store, and pay-per-result data on Apify [4][5].

Sources

Featured on AgentBuiltIs this your business? Get your badge.

Add the badge to your site or README. It links back to this case, and updates itself.

Get your badge Share on X

HTML

<a href="https://agent-built.vercel.app/biz/cairnware?ref=badge" target="_blank" rel="noopener"><img src="https://agent-built.vercel.app/badge/cairnware.svg" alt="Cairnware (Claude-run software studio) is featured on AgentBuilt" height="54" /></a>

Markdown

[![Featured on AgentBuilt](https://agent-built.vercel.app/badge/cairnware.svg)](https://agent-built.vercel.app/biz/cairnware?ref=badge)

#claude#ios-app#apify#seo-tools#build-in-public#zero-revenue

Related cases

SaaSAnethoth (4 agent-built SaaS products)$0 (2 signups in 10 days)

SaaSThumbnailCreator (growth run by a Devin agent)$25.00 collected from 1 paying customer (Stripe pack; $0 via x402), as of Oct 21, 2026

SaaSMintPage (SaaS built and run by an AI agent)$0 after 4 days (0 paying customers; public metrics page)

Info productsZELVEN (Claude on a Raspberry Pi with €50)€0 after 48 work sessions, 0 sales (Oct 16, 2026)

Numbers come from the operators or the press and haven't been audited. Anything missing is marked "not disclosed". Methodology · Report a correction.