# Car Identifier: CarSnap Aiolum story: https://aiolum.com/story/car-identifier/ · added 2026-11-03 · updated 2026-11-03 **In short.** Car Identifier: CarSnap (AI car identification and car-spotting app for iPhone, Canada): US$576 a month recurring revenue, November 2026. Evidence grade A: disclosed by a company or a filing, or verified by a payment processor. Team: 1 founder, Marton Zeisler, who wrote in 2024 that he built the app and the backend alone; headcount not published. Launched 9 September 2024; still running, revenue flat. AI car identification and car-spotting app for iPhone · Canada · AI-native · solo founder · still running, revenue flat · group: Photos and portraits - Outcome: US$576 a month recurring revenue, November 2026 (evidence grade of the headline number: A; headline in US dollars: 6912) - Launch: 9 September 2024; months from launch to the outcome: 26 - Team: 1 founder, Marton Zeisler, who wrote in 2024 that he built the app and the backend alone; headcount not published - Outside funding: Not stated in the sources - Main factor: Niche - Meters, 1 to 4: evidence strength 3, copyable pattern 3, capital needed 1 - Tags: AI-native, Solo founder, Bootstrapped, Running, B2C, Automotive Links: - App Store: https://apps.apple.com/us/app/id6651825365 - Verified revenue on TrustMRR: https://trustmrr.com/startup/car-identifier **The product.** An iPhone app that names a car from a photo. The user takes a picture, picks one from the photo library or shares one from Photos, and the app returns the make, model and year with horsepower, acceleration, top speed, an estimated market value, production numbers and a rarity score. Every identified car goes into a personal garage, is pinned on a map where it was spotted, and unlocks the logo of its brand as a badge. A Pro subscription removes the limits of the free version. **Why it worked.** Naming an unknown car from a photo used to need a computer-vision team. In 2024 one developer could send the photo to a large AI model with an instruction and get structured data back, so he built and published the app in five weeks for a RevenueCat app competition. A payment-data tracker shows that it has earned about US$500 to US$800 a month for most of the past year, with US$15,707 in total since release. Evidence grades: A = disclosed by a public company, a filing or a payment processor, or verified by one; B = two or more independent reports agree, or a journalist confirmed it; C = a founder claim that no one else verified; D = reported, not verified: one founder claim or a third-party estimate with no second source; calc = computed on this page from graded facts. ## Summary Car Identifier: CarSnap is an iPhone app for car enthusiasts. The user photographs a car, and a multimodal AI model, a language model that also accepts images as input, says which car it is. The app then shows its specifications, value and rarity and saves it to the user's collection. Marton Zeisler, the developer and seller named on the App Store, released version 1.0 on 9 September 2024. He wrote on Devpost that he built the iPhone app in Swift and the server part in Python by himself, and that the server sends each photo to Anthropic's Claude model. The revenue figures come from TrustMRR, a site that reads them from the app's RevenueCat account. RevenueCat is a service that records App Store subscription payments for app makers. On 3 November 2026 TrustMRR showed MRR (monthly recurring revenue: the value of all active subscriptions, counted per month) of US$576 from 105 active subscriptions, US$505 of revenue in the last 30 days, US$7,106 in the last 12 months and US$15,707 since release. These figures are grade A. The headline run rate on this page, US$6,912 a year, is the MRR times 12. This is a small business: the monthly revenue in the public timeline stayed between US$303 and US$806 from December 2025 to October 2026. It is shown here because it is fully verified and because the method is easy to copy. The page also shows earlier companies that tried to recognise objects with a phone camera or to build a business on collectors and car buyers, and why they failed. ## Pricing Car Identifier prices in November 2026, from the App Store in-app purchase list and the TrustMRR profile (grade B and C) | Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit | |---|---|---|---| | Free | 0 | 0 | limited features, as the terms say | | Car Identifier Pro (Monthly) | | 9.99 | all features | | Car Identifier Pro (Yearly) | 3.33 | | US$39.99 a year | | Car Identifier Pro (Yearly), second price | 1.67 | | US$19.99 a year | | Car Identifier Pro | | | US$4.99; period not stated on the App Store | | Unlimited | | | US$29.99; listed twice, period not stated | | Car Identifier Pro (Lifetime) | 39.99 one-time | | founder-written TrustMRR profile only | All purchases go through Apple. The App Store lists two yearly prices and two Unlimited items without saying which users see which, and lists Car Identifier Pro at US$4.99 without a period. The founder-written part of the TrustMRR profile also names a lifetime purchase at US$39.99, which the App Store list did not show on 3 November 2026. The app's terms describe a free version with limited features; reviewers wrote that the free version led straight to the paywall. ## How much revenue does Car Identifier: CarSnap make? The numbers, graded A one-developer iPhone app with US$576 of verified monthly recurring revenue and US$15,707 of verified revenue in 26 months | Fact | Value | Grade | Note | Sources | |---|---|---|---|---| | Monthly recurring revenue, 3 November 2026 | US$576, from 105 active subscriptions | A | TrustMRR: "Current MRR: $576", "Current active subscriptions: 105", "Verified payment provider API source: RevenueCat (API key)", revenue last synced 2026-11-03. MRR growth in the last 30 days: +16.2%. | [1] | | Run rate from the MRR | US$6,912 a year | calc | Calculated: US$576 x 12. This is the outcomeUsd on this page. | [1] | | MRR per active subscription | about US$5.49 | calc | Calculated: US$576 / 105. This lies between the yearly plans counted per month (US$1.67 and US$3.33) and the monthly plan (US$9.99), so the subscribers are a mix of yearly and monthly plans. | [1, 2] | | Revenue in the last 30 days, to 3 November 2026 | US$505 | A | TrustMRR, with "Last 30 days revenue growth: +24.6%" and US$170 in the last 7 days. | [1] | | Revenue in the last 12 months | US$7,106 | A | TrustMRR. Calculated: about US$592 a month on average. | [1] | | Revenue since release | US$15,707 | A | TrustMRR, all time. Calculated: US$8,601 of it came before the last 12 months, in the first 14 months after release. | [1] | | Monthly revenue, December 2025 to October 2026 | from US$303 (September 2026) to US$806 (August 2026) | A | TrustMRR monthly timeline: US$515, 518, 449, 680, 701, 697, 687, 681, 806, 303 and 604. Calculated: October 2026 is 15% below the March to August average of US$709 and 2% above the 12-month average, so this page calls the trend flat. | [1] | | Profit margin, last 30 days | 85% | C | TrustMRR snapshot: "Last 30 days profit margin: 85.0%". The public profile does not say how the costs were entered, so this page does not grade it A. | [1] | | Release date | 9 September 2024 | B | App Store version history, version 1.0, and the TrustMRR founded date 2024-09-09. | [2, 1] | | Time from release to the MRR figure | 26 months | calc | Calculated: September 2024 to November 2026. | [1, 2] | | Team | 1 developer | C | Devpost, 2024: "I've worked on this app by myself, I've developed both the app in SwiftUI and the backend in Python." The App Store names Marton Zeisler as developer and seller. No source gives a later headcount. | [3, 2] | | Country | Canada | C | Founder-written part of the TrustMRR profile: "Country: CA". No other source states it. | [1] | | AI model | Anthropic's Claude, called from a Python backend on Firebase Functions | C | Devpost: "The app uses the Anthropic AI to recognize cars in photos." The TrustMRR tech stack lists Anthropic, Firebase, Google Cloud, Python and RevenueCat. | [3, 1] | | Time to build the first version | 5 weeks | C | Devpost: "I managed to build this app and published it within 5 weeks thanks to this Revenue Cat competition" (RevenueCat Ship-a-ton). | [3] | | App Store ratings, 3 November 2026 | 4.2 from 244 ratings in the US store | B | App Store page. TrustMRR's weekly copy of Apple's data on 26 October 2026 showed 4.2 from 241 ratings. | [2, 1] | | Age rating | 4+ | B | App Store. The description is written for "car lovers" at car shows and meets; no page reviewed is aimed at children. | [2] | | Outside funding | not stated | C | No source mentions investors. The tag Bootstrapped reflects that no funding is reported, not a statement by the developer. | [1, 3] | How to read this table: a grade A value was read by TrustMRR from the app's RevenueCat account. A grade B value is a record that Apple shows on the App Store. A grade C value is a statement by the developer, on Devpost, in the app's terms, privacy policy or release notes, or in the founder-written part of the TrustMRR profile. A "calc" value was computed on this page from the rows above it. ## When did Car Identifier: CarSnap launch, and how did it grow? Timeline From a five-week competition entry to a rebuilt version 2.0 two years later. Each entry carries its grade. - **9 September 2024** (grade B [2, 1]): Version 1.0 is released on the App Store for iPhone. - **17 September 2024** (grade C [2]): Version 1.2 brings a redesigned car detail view, one of six updates in the first four weeks. - **19 September 2024** (grade C [3]): The developer posts the app on Devpost as his entry to RevenueCat's Ship-a-ton competition, built and published in five weeks. He plans short videos on TikTok and later paid ads on Instagram and Facebook. - **March 2025** (grade C [2]): Versions 1.12 and 1.13 add a search bar for car brands and previously identified models. - **December 2025** (grade A [1]): First month in TrustMRR's public timeline: US$515 of revenue. - **August 2026** (grade A [1]): Best month in the public timeline: US$806. September follows at US$303. - **22 October 2026** (grade C [2]): Version 2.0, "rebuilt from the ground up": a map of spotted cars, insights, owner opinions, videos and a share option from Photos. Five more updates follow by 31 October. - **3 November 2026** (grade A [1]): TrustMRR shows US$576 of MRR from 105 active subscriptions and US$15,707 of revenue since release. ## Why did Car Identifier: CarSnap work? ### One niche, one clear question Car enthusiasts often see a car they cannot name. The app answers that one question from a photo and adds the numbers enthusiasts compare: power, acceleration, value, how many were made and how rare the car is. ### A large AI model instead of a vision team The developer did not train an image model. The app sends the photo to his own server, which sends it to Claude with an instruction, checks that the answer has a fixed structure, and returns it to the app. He wrote that most of the work was the prompt that keeps the answer format consistent. ### Collecting keeps users coming back Each car goes into a garage, onto a map and into collections, and each new brand unlocks a badge from more than 400 car brands. An insights screen shows the fastest, rarest and most valuable finds. These features give a reason to open the app again after the first identification. ### Paid plans through Apple, tracked by RevenueCat Free users meet limits, and Pro unlocks everything for US$9.99 a month or a yearly price. Each identification is a paid AI call, so the limits also cap the cost of free users. The TrustMRR profile reports an 85% profit margin for the last 30 days. ## How it was built These details come from the developer's Devpost page, the App Store page, the app's terms and privacy policy, and the tech stack on the TrustMRR profile. They are grade B and C. - A native iPhone app in Swift and SwiftUI, about 49.8 MB, requiring iOS 17 or later. - A Python backend on Firebase Functions acts as a proxy to the Anthropic API. The API keys are kept on the server in Google Cloud Secret Manager, not in the app, so that nobody can extract them from the app or from its network traffic. - The backend turns each photo into a prompt for Claude and formats the answer before sending it back to the app. - No accounts: the privacy policy says the app has no login and collects anonymous device-level analytics with Firebase Analytics and Firebase Crashlytics. - Subscriptions are sold through Apple and managed with RevenueCat, which also gives TrustMRR its read-only revenue data. - The terms say that both the free and the Pro versions have server-side limits to prevent abuse and overload. Sources: [3, 2, 6, 5, 1] ## Who tried this before Car Identifier: CarSnap, and why did they fail? Four companies in our database of failed startups tried to recognise objects with a phone camera, to serve collectors, or to give car buyers better information. The values below are from that database and are not independently verified. The failed startups come from our database of dead startups, retold here in our own words; Its entries are AI-assisted summaries of public sources, so they carry no evidence grade. ### Moodstocks (2008 to 2016; funding US$5M; cause: Competition) - Promise: Instant recognition of objects with a phone camera. - Why it died: Could not stay ahead of Google's image recognition and was bought and absorbed by Google. - Record: ### Blippar (2011 to 2018; funding US$130M; cause: Unit Economics) - Promise: Point the phone at an object and see videos and 3D content about it. - Why it died: Brands paid only if users came, and users came only for content; the economics never worked. - Record: ### HobbyDB (2014 to 2022; funding US$5.2M; cause: Competition) - Promise: One database to catalog every collectible ever made, with a marketplace. - Why it died: Spent its funding competing with eBay, while hobbyists already gathered in Facebook groups. - Record: ### Carsabi (2012 to 2015; funding US$1.2M; cause: Competition) - Promise: A used-car search engine with price analytics for buyers. - Why it died: AutoTrader had more money and better technology, and Carsabi relied on organic growth that did not come. - Record: ## What changed Moodstocks and Blippar had to build their own image recognition, which cost millions and which Google did better. In 2024 a single developer could send a photo to a large AI model and get the make, model, year and specifications back, paying per call. HobbyDB and Carsabi needed large audiences and marketplaces to earn money; CarSnap charges the user directly through the App Store and needs no second side of a market. What has not changed is competition. The AI models that make CarSnap possible are open to every developer, and other car identification apps sell the same function. The revenue has stayed at a few hundred US dollars a month for a year, which is what a niche app without a growth channel can look like. ## Can one person copy Car Identifier: CarSnap now? **Verdict: Yes, the method is open to anyone. The revenue is small.** A photo-identification app for one hobby, built by one developer in five weeks on a large AI model, has earned US$15,707 in 26 months, verified from its payment records. It shows that the build is easy; the hard part is getting enough users to pay. 1. Pick a hobby in which people often want to name what they see, and in which the details (specifications, rarity, value) are what they talk about. 2. Send the photo to a large multimodal model from your own server, keep the API key there, and force a fixed answer structure. 3. Add collecting features, such as a garage, a map and badges, so that users return. 4. Limit free use, because each identification is a paid AI call, and say clearly what the free version includes. **What is still open.** Other collecting hobbies with an identification problem: aircraft, trains, watches, sneakers, coins and stamps. Some already have apps, so the choice of niche and the marketing matter more than the AI. ## What could go wrong for Car Identifier: CarSnap? - **Small revenue, no clear growth.** The MRR was US$576 on 3 November 2026. Monthly revenue in the public timeline stayed between US$303 and US$806 from December 2025 to October 2026, and the first 14 months brought about as much per month as the last 12. The last 30 days grew 24.6% after a weak September, and version 2.0 shipped on 22 October 2026; whether that changes the trend is not yet visible. - **Complaints about the free version.** The App Store calls the app "Free · In‑App Purchases" and the terms describe a free version with limits. Two reviewers wrote otherwise: "you cannot identify even a single car without signing up for a subscription (writing this in May 2025)" and "I can't do a single thing unless I pay". A buyer should expect a paywall before the first identification. - **Rated 4+, aimed at car enthusiasts.** The App Store rating is 4+, the default for a reference app. The description speaks to "car lovers" at car shows and cars-and-coffee meets, and no page reviewed markets the app to children. The app stores each car's location on a map; the privacy policy says it has no accounts and collects anonymous analytics. - **Few sources outside the app.** Besides the payment data, the sources are the developer's own pages: the App Store listing, the Devpost entry, the terms and the privacy policy. The country (Canada) and the 85% margin come from the founder-written part of the TrustMRR profile. Devpost also names him as the developer of another app, Hashtag Guru; the App Store developer page listed only CarSnap on 3 November 2026. ## Sources 1. TrustMRR: Car Identifier (markdown profile with RevenueCat-verified metrics, revenue synced 3 November 2026) — https://trustmrr.com/startup/car-identifier.md (grade A). Read the markdown profile, downloaded 3 November 2026. Source for MRR, active subscriptions, revenue by period and month, the founded date and the founder-written country, prices and tech stack. 2. App Store: Car Identifier: CarSnap, by Marton Zeisler — https://apps.apple.com/us/app/car-identifier-carsnap/id6651825365 (grade B). Read in full, downloaded 3 November 2026: description, ratings, reviews, version history, age rating, size and in-app purchase list. 3. Devpost: Car Identifier: CarSnap (RevenueCat Ship-a-ton entry) — https://devpost.com/software/car-identifier-carsnap (grade C). Read in full. The developer's own account of how he built the app, the Claude backend, the five weeks and his marketing plans. 4. App Store developer page: Marton Zeisler — https://apps.apple.com/us/developer/marton-zeisler/id1476767426 (grade B). Read in full. Lists one app, Car Identifier: CarSnap. 5. Car Identifier: Terms and Conditions (Google Sites) — https://sites.google.com/view/car-identifier-termsconditions/home (grade C). Read in full. Source for the free and Pro versions and the server-side limits. 6. Car Identifier: Privacy Policy (Google Sites) — https://sites.google.com/view/caridentifier-privacy-policy/home (grade C). Read in full. Source for the absence of accounts, Firebase Analytics and Crashlytics, and RevenueCat. 7. Database of failed startups, local export — /method/#dead (grade C). Local export. AI-written summaries, not independently verified. Source for the failed predecessors.