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 ›
Live ranking · 73 cases · updated daily

Who's actually making money with AI agents?

Every agent-built business in the library, ranked by the revenue its builder reported, next to what it cost to run. Sourced where possible; every estimate is flagged.

$9Median revenue (incl. $0)
25/65Made exactly $0
44/65Made under $100
$31.2MTop earner · Clanker
22PolsiaPolsia agents$500Kmonthly · ~$500K/month within 3 months, founder-reportedSelf-reportedcost n/d
11ClankerClanker (autonomous bot)$31.2Mtotal · all-time protocol fees, DefiLlama Jul 2026Third-partycost n/d
33Felix (The Masinov Company)OpenClaw (Claude)$195Ktotal · $100,570 Stripe + $94,973 ETH, as of early Mar 2026Self-reportedcost $1,500 Src
The money spectrum

Where every business lands

65 cases with disclosed revenue on one log scale. Hover a dot for the number; click for the sourced case.

Money spectrum of AI agent businesses65 cases with disclosed revenue on a log scale from $0 to $31.2M. 25 sit at exactly $0; the median is $9.$1$10$100$1,000$10K$100K$1.0M$10.0M$0median $9
25 made exactly $0median $9top $31.2M · Clankerlog scale · periods differ · method
73/73
  1. 11
    ClankerClanker (autonomous bot) · Crypto / x402
    $31.2Mtotal
    all-time protocol fees, DefiLlama Jul 2026
    Costn/d Not disclosed and not credibly estimable for a protocol this size.
    Net · ROIn/d
    Third-party
  2. 22
    PolsiaPolsia agents · SaaS / app
    $500Kmonthly
    ~$500K/month within 3 months, founder-reported
    Costn/d Not disclosed and not credibly estimable for a company this size.
    Net · ROIn/d
    Self-reported
  3. 33
    Felix (The Masinov Company)OpenClaw (Claude) · Digital product
    $195Ktotal
    $100,570 Stripe + $94,973 ETH, as of early Mar 2026
    Cost$1,500 Src$1,500 startup cost [1]. Running cost (~$1,500/month, secondary coverage) is not totalled here.
    Net · ROI+$194K+12,900%
    Self-reported
  4. 4
    DropmagicCursor + Claude Code (Hermes for v2) · SaaS / app
    $120KMRR
    ~$120K MRR self-reported Oct 6 2026; earlier €100K/$117K MRR and Apr €106K/~$125K
    Cost$200 Est.Claude Max-class ~$200/mo assumed for heavy Claude Code usage (CASE_TEMPLATE); product LLM/image bill (~$1.5K/mo in Starter Story) is separate COGS not counted as aiOpCost here [4].
    Net · ROI+$120K+59,900%
    Self-reported
  5. 5
    ClawBox + Mike (OpenClaw)OpenClaw (Mike) + Claude Sonnet / Ollama 7B · E-commerce / merch
    $88K49 days
    €76K+ at 1.16; 150+ orders; self-reported DEV Mar 20 2026
    Cost$5 Src~$5/month Claude API after model routing [1]; device retail €549 is product COGS/price, not monthly op cost.
    Net · ROI+$88K+1,763,100%
    Self-reported
  6. 6
    ChatSEOChatSEO's SEO agents (the product) · SaaS / app
    $33KMRR
    Stripe-verified via TrustMRR; €29/month plans
    Costn/d Not disclosed beyond zero ad spend; team and model costs for 671 subscriptions can't be credibly estimated.
    Net · ROIn/d
    Third-party
  7. 7
    Andon Market (run by Luna)Claude (Luna) · Physical retail
    $15Ktotal
    ~$15K sales; not yet profitable
    Cost$48K SrcDerived: $100K start − $67K bank balance + ~$15K sales ≈ $48K spent (lease, staff, stock).
    Net · ROI−$33K−69%
    Self-reported
  8. 8
    Claw MartOpenClaw (Claude) · Marketplace
    $14Ktotal
    marketplace revenue
    Cost$200 Est.One month of Claude Max ($200) assumed for the OpenClaw agent; duration and other costs not disclosed.
    Net · ROI+$14K+6,900%
    Self-reported
  9. 9
    Helena / Project FSD (Enrich Labs' AI-run Shopify store)Helena (Enrich Labs' autonomous marketing agent; underlying model not disclosed) · E-commerce / merch
    $12Ktotal
    $11,532.17 Shopify sales, 170 orders, Jun 1 to Sep 21, 2026; revenue not profit, Phase 1 unprofitable
    Cost$12K Est.Partly sourced: the hero band's $6,280.81 Meta spend and ~$2,000 product cost are stated [1]. The other 25 tests' spend isn't totalled, so we add 25 × ~$150 (the policy's per-test cap) = $3,750 as an assumption. Product costs for the other products and Helena's own platform cost are not included.
    Net · ROI−$499−4%
    Self-reported
  10. 10
    BrandAxisMac mini agents + Gemini · SaaS / app
    $10Kmonthly
    self-reported, Starter Story Jun 2026
    Cost$5,000 Src~$5,000 on tools, tech equipment, R&D and company setup [1]; running and model costs not disclosed.
    Net · ROI+$5,000+100%
    Self-reported
  11. 11
    Andon Café (run by Mona)Gemini 3.1 Pro, then GPT-5.5 (Mona) · Physical retail
    $9,000~2 months
    Gemini period sales
    Cost$38K Src~$38K spent in the Gemini period [3].
    Net · ROI−$29K−76%
    Self-reported
  12. 12
    Ron (Robbie Houston's OpenClaw agent)OpenClaw (Ron) · Content / media
    $8,374MRR
    270 members in 13 days
    Cost$2,374 SrcDerived monthly: $8,374 MRR − ~$6,000/month reported net profit. $100 starting budget [1].
    Net · ROI+$6,000+253%
    Self-reported
  13. 13
    Andon FM (four AI-run radio stations)Claude, Gemini, GPT, Grok · Content / media
    $5,700total
    4 stations
    Cost$5,700 SrcExpenses matched income every day [1].
    Net · ROI$0+0%
    Self-reported
  14. 14
    Vendings and Stuff (Project Vend)Claude (Claudius) · Physical retail
    $2,649quarter-to-date
    phase 2 revenue vs $15,000 target
    Cost$1,000 SrcPhase 1 started with about $1,000 [3]; no clean cost total disclosed.
    Net · ROI+$1,649+165%
    Self-reported
  15. 15
    AgentUseCases.aiGrok Bot · Content / media
    $1,100total
    2 sponsors
    Cost$30 Est.Grok Bot subscription assumed at ~$30/month for ~1 month; Vercel, GitHub, Resend and Stripe free tiers (stated in case).
    Net · ROI+$1,070+3,567%
    Self-reported
  16. 16
    Jarvis (IDIOGEN's autonomous CEO agent)Jarvis (OpenClaw, Claude Opus 4.6) · Service / agency
    $1,02239 days
    4 customers
    Cost$341 Est.$50 budget + Upload Post $24/month × 1.3 (sourced) + Claude Max $200/month × 1.3 assumed for model spend (unpublished).
    Net · ROI+$681+200%
    Self-reported
  17. 17
    EarendelEarendel (autonomous agent) · Digital product
    $8807 days
    net, 26 paid sales
    Cost$50 Src$50 starting budget [1]; other costs not disclosed.
    Net · ROI+$830+1,661%
    Self-reported
  18. 18
    Larry (Snugly & Liply growth agent)OpenClaw (Claude) · SaaS / app
    $714MRR
    app subscriptions
    Cost$30 Est.~$0.50 API per post (sourced) × assumed 2 posts/day × 30 days; hardware was a spare PC.
    Net · ROI+$684+2,280%
    Self-reported
  19. 19
    AI Village merch store (Claude Opus 4)Claude Opus 4 · E-commerce / merch
    $564total
    24 orders, $126 profit
    Cost$438 SrcDerived: $564 revenue − $126 reported profit.
    Net · ROI+$126+29%
    Self-reported
  20. 20
    Rosalinda Solana: OpenClaw operator selling setup service and PDFs (rosasolana)Rosalinda Solana (OpenClaw orchestrator + builder + amplifier agents; Claude Sonnet/Opus) · Service / agency
    $429total
    $429.12 from 13 sales, all time (Mar 16, 2026); $377.89 from 11 OpenClaw setup jobs
    Cost$90 Est.Model costs not disclosed; it runs on Anthropic's API [7]. Assumes an always-on API heartbeat loop at ~$1.50/day for about two months (late January to March 2026) = ~$90.
    Net · ROI+$339+377%
    Self-reported
  21. 21
    HustleGPT / Green Gadget GuruGPT-4 (directing a human) · E-commerce / merch
    $130total
    ~$130, mostly promoted tweets; the viral $7.7K–7.8K was investment
    Cost$100 Src$100 starting budget given to GPT-4; day-1 spend from it was $8.16 domain + $28 hosting + $40 ads allocated [2][3].
    Net · ROI+$30+30%
    Self-reported
  22. 22
    BugBasherDevin · Service / agency
    $754 weeks
    Cost$3,000 SrcAbout $3,000 in costs [1].
    Net · ROI−$2,925−97%
    Self-reported
  23. 23
    NanoCorp origin experiment ($10 course agent)Autonomous agent (not specified) · Digital product
    $706 days
    then Stripe shut it down
    Cost$10 Src$10 budget [1].
    Net · ROI+$60+600%
    Self-reported
  24. 24
    Olivia (Olivia Craft)Olivia Craft (self-described autonomous agent) · Digital product
    $5410 weeks
    2 sales
    Cost$50 Est.$20/month model plan assumed × ~2.5 months.
    Net · ROI+$4+8%
    Self-reported
  25. 25
    Claw (Kinvero / PebbLearn)OpenClaw (Claude) · Service / agency
    $4730 days
    Cost$411 Src$411/month overhead [1].
    Net · ROI−$364−89%
    Self-reported
  26. 26
    Operation TalonOpenClaw (Talon) · Digital product
    $298 days
    first sale
    Cost$200 Src~$200 budget (Claude Max) [1].
    Net · ROI−$171−85%
    Self-reported
  27. 27
    ThumbnailCreator (growth run by a Devin agent)Devin (AI coding agent, running Claude) · SaaS / app
    $25total
    $25.00 from 1 Stripe pack buyer since the agent took over growth (Sep 26 to Oct 21, 2026)
    Cost$33 Est.Sourced: ~$5.90 growth budget [1] and ~$7 of model costs used by bot accounts [3]. Assumed: $20 for the agent's own runtime (one weekly Devin run; Devin's cost is not disclosed, standard $20/month plan-equivalent).
    Net · ROI−$8−24%
    Self-reported
  28. 28
    Earn My Keep (Spinotto, an AI with 30 days to earn €100)Spinotto (AI agent; model not named on the site) · Digital product
    $24.7523 days
    €21.34 from sales and sponsors (EUR at 1.16); the ledger's €65.29 also includes a €43.95 bet the human placed
    Cost$154 Src€100 AI subscription for October + €33.08 of ledger spending (Etsy, Amazon ads, AI services, fees) through day 23 = €133.08, at 1.16 USD/EUR ≈ $154 [3].
    Net · ROI−$129−84%
    Self-reported
  29. 29
    Acrid AutomationClaude Code (Acrid) · Digital product
    $17total
    Cost$200 Est.One month of Claude Max ($200) assumed for Claude Code; nothing disclosed.
    Net · ROI−$183−91%
    Self-reported
  30. 30
    clodfarm → Towers (Opus 5.5 business farm)clodfarm (Claude Opus 5.5 agents via Claude Code) · SaaS / app
    $13total
    $13 in ~3 weeks vs 'over $500' spent; Towers' 63 companies $13 combined (Oct 21, 2026)
    Cost$500 Src'We wasted over 500$' (founder, Oct 21 update) [2]. A lower bound, covering the whole experiment; the split between three Claude subscriptions, AWS and ads isn't disclosed.
    Net · ROI−$487−97%
    Self-reported
  31. 31
    ONE HOUR (1h-money)Claude · Digital product
    $12.7657 minutes
    €11 gross (€10.15 net), EUR at 1.16
    Cost$1 Est.One hour of a $20/month Claude plan, rounded up to $1.
    Net · ROI+$11.76+1,176%
    Self-reported
  32. 32
    from100to200 (Claude doubling €100)Claude · Digital product
    $9.58total
    €8.26 net from 1 sale, EUR at 1.16
    Cost$11.83 Src€10.20 domain, the only ledger outflow [1].
    Net · ROI−$2.25−19%
    Self-reported
  33. 33
    AI Tools Insider: Money-Agent KitAutonomous agent (self-reported) · Digital product
    $975+ days
    first sale
    Cost$48 Est.950+ agent runs × assumed ~$0.05 API per run.
    Net · ROI−$39−81%
    Self-reported
  34. 34
    Agent Receipts (Matt Busel's AI distribution team)Claude-based agent team (6 roles) · SaaS / app
    $7.74~1 week
    tracked revenue
    Cost$50 Est.Claude Max ($200/month) assumed, prorated to ~1 week.
    Net · ROI−$42.26−85%
    Self-reported
  35. 35
    AI Hustler (Apify scrapers)Claude (CEO + worker sessions) · SaaS / app
    $6.7441 days
    200+ runs
    Cost$273 Est.Claude Max ($200/month) assumed × 41 days ≈ $273.
    Net · ROI−$266−98%
    Self-reported
  36. 36
    Swift Bot Labs (AI agent paying its own $6.67/day)Swift Bot Labs agent (model not disclosed) · Service / agency
    $6.67total
    $6.67 from a video + two reports paid on-chain (Oct 21, 2026); equals one day of running cost
    Cost$26.68 Src$6.67/day running cost [1] × 4 days (Oct 18–21, 2026) = $26.68, our arithmetic from the sourced daily rate. Model/compute split not disclosed.
    Net · ROI−$20.01−75%
    Self-reported
  37. 37
    JeezAutonomous agent (Claude-based) · Digital product
    $4.9930 days
    one customer
    Cost$200 SrcDerived: ~$40 spent per $1 earned [1] × $4.99.
    Net · ROI−$195−98%
    Self-reported
  38. 38
    Profiterole (autonomous agent team, $3 lifetime)~8 autonomous agents on cron (orchestrator, builder, strategist, analytics, marketer, executor, discovery, thinker, blogger) · SaaS / app
    $3lifetime (to Apr 2026)
    one $3 Buy Me a Coffee tip; $0 via Stripe across 8 packages
    Cost$19 Est.Model and API spend not disclosed. 381 cycles by Mar 21, 2026 x an assumed ~$0.05 of API cost per cycle = ~$19. EC2 hosting isn't included, and later cycles with ~8 agents would add more.
    Net · ROI−$16−84%
    Self-reported
  39. 39
    Krimskrams (run by Hans Krämer)Claude Code + Codex (Hans Krämer) · SaaS / app
    $1.5421 days
    Cost$540 Src$400/month Claude Max + Codex Pro, ~$40 server, ~$100 other expenses [1].
    Net · ROI−$538−100%
    Self-reported
  40. 40
    NomadicForgePrintCo (Grok Bot Etsy)Grok Bot · E-commerce / merch
    $0.99total
    One $0.99 Etsy receipt verified; operator also claimed 5 sales (total $ not disclosed)
    Cost$30 Est.Grok Bot subscription assumed ~$30/month for ~1 month of the challenge; Etsy fees not itemized in sources.
    Net · ROI−$29.01−97%
    Self-reported
  41. 66
    Astra Trading Company (AI-agent-run copy and data studio)Astra CEO agent + marketing, strategy, operations and compliance agents (model not disclosed) · Service / agency
    n/dtotal
    1 paid Speedbot order in week 1; amount not disclosed (offers $5-$49 USDC)
    Cost$45 Est.Model and runtime not disclosed. Five agent roles running daily content, delivery and ops; uses the standard always-on API loop assumption (~$1.50/day, about $45/month). The x402 API itself runs at $0 on Cloudflare's free tier [1].
    Net · ROIn/d
    Undisclosed
  42. 67
    Estelle (AI astrologer)OpenClaw (Darla) + human team rebuild · SaaS / app
    n/dtotal
    50 paying customers in 12h of ads; $ price/MRR not disclosed
    Costn/d Ad spend and weekly inference for 60K texts are not disclosed; OpenClaw/build weekend cost unknown — cannot estimate honestly.
    Net · ROIn/d
    Self-reported
  43. 68
    Jake Essex booking agentGrok Bot (+ Grok Build for site) · Service / agency
    n/d30 days
    30 shows booked (claimed); gig fees not disclosed
    Cost$30 Est.Grok Bot subscription assumed ~$30/month (Jake notes no free tier); site hosting not itemized.
    Net · ROIn/d
    Self-reported · unverified
  44. 69
    Demo FactoryClaude Code (kit for Claude Code) · Digital product
    n/dtotal
    Sales not disclosed; operator still chasing first few (Oct 2026)
    Cost$200 Est.Assumes ~1 month of Claude Max ($200) while packaging and demoing the kit; per-demo API ~$1.45 is usage on top, not totaled.
    Net · ROIn/d
    Undisclosed
  45. 70
    BuildkitClaude Code / Codex / Cursor (consumer of Buildkit skills) · Digital product
    n/dnot disclosed
    Buildkit sales not published; Pixelcave claims 20k+ prior-product customers
    Costn/d No published AI or hosting spend for Buildkit.
    Net · ROIn/d
    Undisclosed
  46. 71
    ShelfSpaceClaude API pipeline · Service / agency
    n/d—
    not disclosed; $497 per audit
    Cost$20 Est.Assumed ~$20 of Claude API usage for building and running audits; Vercel and Stripe have no fixed fee at this scale. Nothing is disclosed.
    Net · ROIn/d
    Undisclosed
  47. 72
    Payhook (built by agent 'Laila')Laila (AI agent) · SaaS / app
    n/d—
    not disclosed
    Cost$20 Est.One month of a $20/month model plan assumed; nothing disclosed.
    Net · ROIn/d
    Undisclosed
  48. 73
    Digital Pin Club + PinsiderGrok Bot · Content / media
    n/d—
    not disclosed
    Cost$30 Est.Grok Bot subscription assumed at ~$30/month for ~1 month.
    Net · ROIn/d
    Undisclosed
$0

Hall of $0

25 businesses that shipped, tried, and earned nothing (yet). Honest data is the point; these teach the most.

SrcCost taken from the case's sources (or simple arithmetic on them).
Est.Our transparent estimate; hover for the basis. Never presented as sourced.
Self-reportedRevenue published by the builder, not audited.
Third-partyRevenue from independent data (e.g. DefiLlama).

Methodology

  • Revenue is the figure the builder or operator published, linked on each case page. It is self-reported unless marked as third-party data. We never estimate revenue; undisclosed revenue shows as n/d and ranks last.
  • Currency: amounts in euros are converted at 1 EUR = 1.16 USD (an approximate October 2026 rate). The original figure stays in the note.
  • Period is shown for each row (total, MRR, monthly, or the experiment window). Rows are ranked by the reported number, so periods are mixed; read the period before comparing.
  • Input cost covers model/API fees, tool subscriptions, ad spend and starting capital as disclosed. Sourced costs are marked Src. For 41 cases without a sourced total we estimate it (Est.) from the stated agent and runtime, using list prices we assume: Claude Max $200/month, Claude or ChatGPT plans $20/month, a Grok Bot subscription ~$30/month, ~$1.50/day for an always-on API loop, and free tiers where the case says so.
  • Net is revenue minus input cost; ROI is net divided by cost. Both mix periods in some rows. Clanker and Polsia publish no costs and are too large to estimate credibly, so their net and ROI are n/d.
  • Revenue bars use a log scale, so $10 and $10M both stay visible.
  • Filters: business type, industry, starting-capital band and AI operating-cost band are stored with each case; AI-cost bands include estimates. Filters and sort live in the URL, so views are shareable.

Spotted a wrong number? Send a correction · Submit your case

FAQ

Which AI agent business made the most money?

Clanker, with $31.2M (total; all-time protocol fees, DefiLlama Jul 2026). The figure comes from third-party data.

How much does a typical AI agent business make?

The median across the 65 cases that disclose revenue is $9. 25 of them made $0, and 44 made under $100.

Are these revenue numbers verified?

No. Almost every figure is self-reported by the builder or operator and linked to its public source. Third-party data (such as DefiLlama for on-chain fees) is labelled as such. We don't audit anyone.

What does input cost include?

Model and API fees, tool subscriptions, ad spend and starting capital, as far as the case discloses them. When a case gives a cost, we use it and mark it Src. When it doesn't, we show a transparent estimate marked Est., with its basis in the tooltip and under the number.

Why do the revenue periods differ?

Builders report differently: lifetime totals, monthly recurring revenue (MRR), or a fixed experiment window. Each row states its period. Rank is by the reported figure in USD, so an MRR case can sit next to a lifetime total.