Photosolve.io: A Homework AI App That Reached 2.5M Users at $0.50 CAC — Sold for $575,000 in Under Two Years

In an era when anyone can use ChatGPT for free, is there still a business in an app that merely solves homework from a photograph? The deal data behind Photosolve.io answers that question with actual numbers. Launched in April 2023 by an operator based in Michigan, the iOS app was listed on the online marketplace Flippa less than two years later and found a buyer at $575,000.

According to Flippa’s published deal report, trailing-twelve-month revenue at the time of sale was $348,849, with profit margins of roughly 86%. Cumulative users stood at 2.5 million, of whom 5,000 were paying subscribers. The most striking figure is customer acquisition cost: approximately $0.50 for US users. Ninety percent of traffic came from organic sources, chiefly viral TikTok activity and App Store recommendations, and the operator’s own time input is listed at around 10 hours a week.

In other words, this business was priced not on how cleverly it solves problems with AI, but on how nearly free it was to acquire users. Against the sale prices, $575,000 sits toward the upper end — and the thing funding that price is not product novelty but a single metric: acquisition cost.

The disclosed numbers

Applying the 86% margin to revenue puts annual profit around $300,000, making the $575,000 price roughly 1.9x. Averaged monthly, profit works out to about $25,000, so the sale price equals roughly 23 months of it. (That multiple is our calculation from the published revenue and margin, Flippa did not publish a multiple itself.)

The listing page carries more traces of the deal. Status: SOLD. Seller and buyer both in the United States. The advising broker was Jared Lauber of Flippa’s North America team. The page logged 2,988 views and 70 watchers. A few thousand people looked, seventy followed it, and one of them bought.

What a $0.50 CAC actually means

An LTV of $42 against a CAC of $0.50 is a ratio of 84 to 1. No business running paid advertising produces that number. And indeed, Flippa’s report states plainly that 90% of traffic was organic: this business did not grow by buying media, it grew on TikTok videos that spread and on App Store search and recommendations. The $0.50 CAC is better read not as evidence of skilled advertising but as the flip side of barely advertising at all.

The same structure is visible in Cal AI, which reached 15 million downloads through TikTok before being acquired. Teenage users show the app they use in short-form video, and that video drives the next download. Student-facing apps have a built-in structure in which the users themselves are the cheapest acquisition channel. Homework apps are the archetype: “point the camera at this and the answer appears” fits comfortably into fifteen seconds.

The other force at work is a very low paid-conversion rate. Five thousand paying users out of 2.5 million is just 0.2%. Normally that is a weakness, but at a $0.50 acquisition cost the arithmetic changes. Two hundred free users cost $100 to acquire, one of them converts, and that one is worth $42 in LTV. On the face of it the math does not close. The business works because the $0.50 figure is specifically for US users and because the vast majority of those 2.5 million arrived at no media cost at all. Try to run the same 0.2% conversion on paid acquisition alone and the economics collapse.

Designing your way out of the “AI wrapper” label

Flippa’s report justifies separating this asset from the generic AI wrappers by pointing at its technical stack: retrieval-augmented generation combining vector and semantic search, an integration with Wolfram Alpha, optimization across multiple LLMs, and a knowledge base built on historical academic data said to handle demanding national exams.

That deserves a discount when you read it: it is the seller’s own description of the asset, re-packaged by the platform as a deal report. Still, it is informative about what the buyer paid for. An app that merely calls an API can be rebuilt over a weekend. Whether you own the machinery that guarantees answer accuracy is presented here as the dividing line for whether an asset trades at this price band.

What is not disclosed also matters. Churn, month-by-month revenue, subscription price points, and how many of the 2.5 million users are active never appear in the report. LTV is given as $42, yet dividing $348,849 of annual revenue by 5,000 paying subscribers gives roughly $70 per subscriber per year. The natural explanation is a different measurement window or the inclusion of free users, but the report defines neither. This is a figure to treat as unverifiable.

The visible weaknesses

This business’s fragility lives in exactly the same place as its strength. Ninety percent organic traffic also means 90% dependence on two mechanisms it does not control: TikTok’s algorithm and App Store ranking. Change either set of criteria and the $0.50 CAC vanishes overnight. A business that buys its traffic experiences that as “costs went up”. A business built on organic traffic experiences it as “traffic stopped.”

There is also the question the homework-app category carries with it. Study support and doing the assignment for the student sit on a continuum, and if institutional attitudes shift, so can store policies and school-level treatment. The seller’s choice to exit quickly is unlikely to be unrelated to that uncertainty, though no reason for the sale is stated in the report.

The limits of the sourcing should be stated too. Every primary figure here comes from Flippa. A marketplace has an incentive to publicize its closings, so there is always a chance that the flattering metrics are the ones selected. No independent reporting or founder-side disclosure could be located. What we verified is that the listing page shows SOLD and that its details are consistent with the deal report.

How far this generalizes

What generalizes is the pricing, not the growth. Annual revenue of $348,849, an 86% margin and roughly 10 hours a week of work produced a $575,000 valuation. As a marker for what an individual can build in two years, that combination is a useful reference point. Feather sold at roughly 3.5x annual revenue. This one at roughly 1.65x. Put side by side, the pair shows that SaaS and apps trade at different multiples, and that assets whose subscription durability is hard to read fetch lower ones.

What does not generalize is the traffic. Going viral on TikTok is closer to an event than a tactic. “Achieve virality” cannot be written into a plan. Build the same homework app today and nothing guarantees 2.5 million users at $0.50 apiece. A more realistic posture is the one taken by the developer who shipped 68 apps over ten years. Keep stepping up to the plate until something connects.

Our reading is this. The lesson is not that “AI apps sell.” It is that an asset with abnormally low acquisition cost commands a price even at modest scale. Revenue of $348,849 is not large for a US small business. It fetched $575,000 because what the buyer was buying was the pipeline through which users arrive without media spend, not the revenue itself. Which also means a business that cannot explain that pipeline will not fetch the same price at the same revenue.