Open callDid an agent build your business?Get a free, sourced case study in the library.Get listed free Submit your case
NYC --:--:--The wallLeaderboardPlaybooksMethodSubmit a case ›
Hall of $0Info products

Build It Together (Thomas Kanze's $200 AI boss)

A $19 pack of 8 weekend AI projects for parents and kids

ClaudeThomas Kanze (@ThomasKanze)Updated 2026-10-30Visit site ↗
kids.jungle.dadBuild It Together (Thomas Kanze's $200 AI boss) website screenshot
Reported revenue · first 9 days (Oct 22–30, 2026)$0Self-reportedcost $19.08 Src
Share on XShare card (PNG)Get the badgeFeatured @ThomasKanze
TL;DR

Thomas Kanze, a dad of three in Costa Rica, gave an AI agent $200 and 30 days and made it his boss: it picks and runs the business, and he only approves spend and posts [1]. Its first pick, a $29 Supabase security report for vibe-coded apps, got 245 landing views and $0 in a week, so it pivoted [3][7]. Since Oct 29 it sells Build It Together, a $19 pack of 8 weekend AI projects for parents and kids, with 0 sales on day one [8][9].

Budget$200 for 30 days[1]
Spent by day 7$19.08[7]
Revenue$0 (to Oct 30)[9]
First idea$29 Supabase report; caught 12 of 12 planted flaws[3]
Landing views, week 1245[7]
Cold emails5 sent, 0 replies[6]
Pack price$19 one-time[8]
AI analysis

Business model analysis

Written by AI on Nov 1, 2026 from the 13 sources at the bottom of this page. Lines with a [n] link are reported facts; lines tagged Inference are the model's own reasoning. No revenue figures are added or estimated here.

VerdictEarly signal
  1. Stopped
  2. No revenue
  3. Loss-making
  4. Early
  5. Earning

Nine days in, two products tried, $0 revenue and $19.08 spent [7][9].

Reported revenue · first 9 days (Oct 22–30, 2026)$0Self-reported
How it makes money

An AI agent with a $200 budget picks and runs a business; it now sells a $19 parent-and-kid AI project pack [1][8].

Who pays
  • Parents of kids aged 7–14 who want a hands-on weekend project [10]
  • Inference Homeschool families: the page pitches one project a week as an 8-week block [10].
Pricing
Build It Together
$19 one-time, 30-day refund [10]
Free sample
First project PDF, no email needed [10]
Earlier idea
Free scan, $29 full report [3]
Cost structure
Domain
$9.08 [2]
Spent by day 7
$19.08 of $200 [7]
AI
Claude plan; price not stated [13]
Growth engine
  • XDaily AI-boss updates: 58 posts, 1,700+ views in a week [9]
  • Cold email5 sent, 0 replies [6]
  • Google adsBlocked: the agent isn't a person [6]
  • Meta adsNext plan; pixel already on the page [9][10]
Moat
  • Inference The founder's dad-in-the-jungle story and audience, not the product. The pack is easy to copy.
Risks
  • Ad platforms need a real, verified person [6]
  • Free AI tool plans may change their limits [10]
  • Inference Short 30-day window to find buyers for a seasonal parent product.
Why it's still early
  1. 01The first idea got views but no buyers [7]
  2. 02Inference A security report sold by an anonymous AI is a hard trust sell; a parent pack sold by a dad fits the founder's story better.

What does Build It Together (Thomas Kanze's $200 AI boss) sell, and to whom?

A $19 pack of 8 weekend AI projects for parents and kids.

First, builders of vibe-coded apps worried their Supabase rules leak user data [3]. Now, parents of kids aged 7 to 14 who want a screen-time afternoon that builds something together [8][10].

How was Build It Together (Thomas Kanze's $200 AI boss) built?

The agent runs on Claude, per Thomas's posts [12][13]. It picked the business, built a Supabase row-level-security checker that caught all 12 flaws planted in a test app, and launched checkout, then broke it when its own safety filter replaced the payment key with '[REDACTED]' [3][4]. For the pivot it produced a 72-page PDF and web version with 8 projects, 12 rainy-day challenges and playable demos [10].

ClaudeGumroadSupabaseMeta pixel

What is the business model?

One-time digital product: $19 through Gumroad with a 30-day refund [10]. The earlier security idea was a free scan plus a $29 full report [3].

Where does the traffic come from?

Daily X posts about the AI boss, 5 cold emails, and an attempt at Google ads that stalled because Google didn't believe the agent was a person [6][9]. The kid-pack page runs a Meta pixel, and the boss wants Meta ads next [9][10].

What revenue has Build It Together (Thomas Kanze's $200 AI boss) reported?

Reported revenue: $0 (to Oct 30, 2026), per @ThomasKanze kid pack update (X) (2026-10-30).

  1. 2026-10-22Gives the agent $200 and 30 days; it's the boss [1]
  2. 2026-10-23Only expense approved: a $9.08 domain [2]
  3. 2026-10-24Picks a Supabase rules checker: free scan, $29 report [3]
  4. 2026-10-25Checkout launched, then broken by a '[REDACTED]' key [4]
  5. 2026-10-27Google ads blocked: not a person; 5 cold emails, 0 replies [6]
  6. 2026-10-28Day 7: $0, $19.08 spent, 245 views; pivot to a kid pack [7]
  7. 2026-10-29Build It Together launches at $19 [8]
  8. 2026-10-300 sales; the boss wants Meta ads [9]

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

Agent
  • Picked both business ideas and set its own deadline rule [3][7]
  • Built the security checker and the kid pack [3][10]
  • Set up checkout [4]
  • Sent cold emails and tried to run Google ads [6]
Human
  • Approves spend and posts only [1]
  • Stepped in when Google wouldn't verify the agent [6]
  • Posts the daily updates on X [1]

What does it cost to run?

$19.08 of the $200 budget spent by day 7, including a $9.08 domain [2][7]. The Claude plan's price isn't stated [13].

What worked and what didn't?

Worked
  • Fast building: a working checker in 3 days and a 72-page pack within a week [3][10]
  • A clear kill rule: 2 paid reports by Wednesday or change the plan [7]
  • Public daily numbers make the story easy to follow [9]
Didn't
  • $0 revenue in the first 9 days [9]
  • 5 cold emails, 0 replies [6]
  • Checkout broke on launch day [4]
  • Ad platforms want a real person [6]
  • Usage limits: 'Claude quit on me at noon' [12]
AI guide

How to copy Build It Together fast

A starter path written by AI from this case. Budgets and timings are our estimates, not the operator's numbers.

DifficultyEasy
BudgetEst.$200 + an AI plan
TimeEst.30 days
Grok Bot templateNone known

Inference Thomas's budget [1] plus a Claude plan (price not stated; we assume $20/month). No public Grok Bot template runs an 'AI as boss' business. For the first idea (checking vibe-coded apps for security holes), Vibe Check (https://grokbotusecases.com/templates/vibe-check/) is the closest public template; it's not what Thomas used.

Tools & agents
Build path5 steps
  1. 1
    Set the rules

    $200, 30 days, real customers only; you approve spend and posts [1].

  2. 2
    Let the agent pick and build

    It shipped a checker in 3 days [3].

  3. 3
    Set a kill rule

    '2 paid reports by Wednesday or change the plan' [7].

  4. 4
    Pivot to your audience

    A parent pack fits a dad's audience [8][11].

  5. 5
    Post every number

    Daily posts with spend, revenue and views [9].

Week 17 days
  1. D1

    Write the budget and approval rules

  2. D2

    Let the agent pick the idea and buy a domain

  3. D3

    Build the first version

  4. D4

    Launch checkout and test a real payment

  5. D5

    Reach 10 likely buyers directly

  6. D6

    Check the kill rule

  7. D7

    Keep, pivot or stop

Pitfalls
  • Test checkout yourself; a safety filter broke the key [4]
  • Ad accounts need a human [6]
  • Expect usage limits mid-day [12]
Key moves from the original case
  • Give the agent a budget, a deadline and a kill rule
  • Let the human approve only spend and public posts
  • Smoke-test checkout with a real payment
  • Plan for a human to pass identity checks on ad and payment platforms

What are the risks and the moat?

Kid-coding guides compete with free tutorials, and the pack depends on free AI tool plans whose limits change [10]. Ads need a verified human [6]. All numbers are Thomas's own posts.

FAQ

Has it made money?

Not yet: $0 as of Oct 30, 2026 [9].

What does the AI boss sell?

A $19 pack of 8 weekend AI projects for parents and kids aged 7–14 [8][10].

Why did it pivot?

Its first product, a $29 Supabase security report, got 245 views and no sales in a week [7].

Which AI is it?

Claude, per Thomas's posts [12][13].

Sources

  1. [1] 'I gave my AI agent $200 and made it my boss' (Oct 22, 2026)↗
  2. [2] Day 2: $9.08 domain, $0 revenue (Oct 23, 2026)↗
  3. [3] Day 3: Supabase rules checker, free scan, $29 report (Oct 24, 2026)↗
  4. [4] Day 4: checkout broken by '[REDACTED]' payment key (Oct 25, 2026)↗
  5. [5] Day 5: checkout works, nobody has used it (Oct 26, 2026)↗
  6. [6] Day 6: Google didn't believe the agent was a person; 5 cold emails, 0 replies (Oct 27, 2026)↗
  7. [7] Day 7: $0 revenue, $19.08 spent, 245 views; pivot to a kid pack (Oct 28, 2026)↗
  8. [8] Launch: Build It Together, $19, 8 projects, ages 7–14 (Oct 29, 2026)↗
  9. [9] Kid pack day 1: 0 sales; the boss wants Meta ads (Oct 30, 2026)↗
  10. [10] Build It Together sales page (checked Nov 1, 2026)↗
  11. [11] jungle.dad: $0 so far toward $10K/month (checked Nov 1, 2026)↗
  12. [12] 'Claude quit on me at noon' (Oct 28, 2026)↗
  13. [13] 'All my money is going towards Claude and no MRR' (Oct 31, 2026)↗
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.

Build It Together (Thomas Kanze's $200 AI boss) is featured on AgentBuilt
Get your badgeShare on X
HTML
<a href="https://agent-built.vercel.app/biz/thomas-kanze-ai-boss?ref=badge" target="_blank" rel="noopener"><img src="https://agent-built.vercel.app/badge/thomas-kanze-ai-boss.svg" alt="Build It Together (Thomas Kanze's $200 AI boss) is featured on AgentBuilt" height="54" /></a>
Markdown
[![Featured on AgentBuilt](https://agent-built.vercel.app/badge/thomas-kanze-ai-boss.svg)](https://agent-built.vercel.app/biz/thomas-kanze-ai-boss?ref=badge)
#claude#ai-boss#kids#education#pivot#build-in-public

Related cases

Info productsfrom100to200 (Claude doubling €100)€8.26 net from 1 sale (ledger at €98.06 of €200)Info productsMoneylab (Claude given $80 to run a business)$10.50 gross across 3 transactions in 216 days, $5.00 from an outside customer (Oct 24, 2026)Info productsZELVEN (Claude on a Raspberry Pi with €50)€0 after 48 work sessions, 0 sales (Oct 16, 2026)Info productsBuilt By Zac (72-hour $100 bet)$0 (16 products, 300 articles)

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