Fiddl.art

Executive Summary

Fiddl.art works because it converted a pricing complaint into a business model. The dominant AI art tools sell subscriptions: Midjourney's cheapest plan is $10 per month and there is no pay-as-you-go option at all (verified: Midjourney pricing docs). Fiddl.art sells points. A casual user who wants ten images this month pays a few dollars once, gets access to 30+ frontier models in one interface, and is never asked to commit. That single pricing inversion, wrapped in a game-style missions economy that lets users earn points instead of buying them, is the wedge. Everything else, the Reddit acquisition engine, the SEO listicle factory, the model-launch news-jacking, exists to put low-commitment buyers in front of that wedge.

The mechanism is specific: aggregate every new image and video model within days of release (Sora support shipped October 2025, Seedream 4.5 coverage December 2025, Nano Banana 2 March 2026, each documented on the company blog), then capture the search and Reddit demand each model launch generates, then monetize that demand with a one-time purchase that converts in a median of 6.5 minutes from signup (self-reported). The company reports it runs no paid ads and spends roughly $2,000 per month on contractors, yielding a reported organic CAC of $6.35 per new cash buyer against a gross LTV of $46.29, a 7.3:1 ratio (self-reported, founder update on the TrustMRR listing). Local TrustMRR Stripe-derived charge data shows $127,242 in cumulative gross charges across 5,428 charges from November 10, 2025 through June 9, 2026, with trailing 30-day volume of $31,289. The business is a low-CAC arbitrage on other companies' model launches, priced for the buyer the subscription incumbents structurally exclude.

The number that matters most is the average charge: $23.44 across all 5,428 TrustMRR-tracked charges, rising from roughly $18 in December 2025 to $34 in May 2026 (local TrustMRR data). Basket size is growing while charge counts fell from 1,323 in March to 838 in May. The buyers are getting fewer and bigger. That is the early signature of either a maturing power-user base or a topping-out casual funnel; the next two quarters decide which. Our view, by a slight majority, is the former, because the company's own cohort data shows 80% of users who spend 5,000+ points convert to paying (self-reported).

What this teardown is not: a story about proprietary AI. Fiddl.art trains no frontier models, owns no model IP, and runs on Vue, Node, OpenAI APIs, Google Cloud, and DigitalOcean (third-party: TrustMRR tech stack). It is a distribution and pricing business sitting on top of commodity inference. That is precisely why it is copyable, and why the operator lessons below are worth the read. The company is currently listed for sale at $1,150,000, roughly 3.3x trailing revenue, with 11 offers received (third-party: TrustMRR listing, June 2026). The founder's own listing note concedes that retention and predictability are "not yet fully proven." Read this as a playbook for the first $130k of gross charges, not a blueprint for a durable moat.