AppAlchemy
AppAlchemy is two AI products run by one founder. The first, at design.appalchemy.ai, draws the screens of a mobile app from a text description or a reference image; the user asks for changes in a chat and exports the result as HTML or a share link. The second, launched in July 2026 at appalchemy.ai, is a no-code app builder, a service that makes software for people who do not write code: the user describes an app, and the service writes React Native code, a framework in which one code base runs as both an iOS and an Android app, and shows it live on the user's phone. Both products sell monthly credits, units of AI work that each request uses up, and both bill through one Stripe account.
Diego Roshardt, who signs as "Dro" and gives Miami as his location on X, says he had the idea after he paid a designer on Upwork hundreds of dollars for an app design and more for every change. He built the first version in about two weeks, started marketing it on Reddit right away, and found his paying customers there while his X account had 70 to 115 followers. He posted MRR, monthly recurring revenue, the money that subscriptions bring in each month, of US$3,219 on 23 April 2025, US$10,000 on 20 May and US$17,000 on 14 June. TrustMRR, a site that reads a startup's revenue directly from its payment processor, the company that charges the customers' cards, shows Stripe revenue of US$18,298 in October 2025, the best month in its public data, and US$243,652 in total by 14 October 2026.
After October 2025, revenue was lower than in the month before in 9 of 11 months. September 2026 brought US$5,860, 68% below the best month, and on 14 October 2026 MRR was US$5,408 from 193 active subscriptions. The founder writes that he stopped marketing in early 2026 and that he is burned out. He has listed the business for sale on TrustMRR at US$27,500, down from US$75,000 in March 2026. The revenue figures on this page come from Stripe and are grade A; what the founder says about customers, costs and reasons is grade C.
The numbers, graded
Each row shows the evidence grade and the source numbers from the list at the end of this page. The grades are explained in How Aiolum verifies.
| Fact | Value | Grade | How we know | Sources |
|---|---|---|---|---|
| Revenue in the best month on record, October 2025 | US$18,298 | A | Stripe data through TrustMRR, in the archived profile of 14 September 2026. The public monthly data start in October 2025, so an earlier month could have been higher. | [2] |
| Run rate in that month | about US$219,576 a year | calc | Calculated: US$18,298 times 12. Revenue has since fallen, so this was a peak rate, not today's. | [2] |
| Months from founding to the best month | 8 | calc | Calculated: February 2025 to October 2025. The founding date, 11 February 2025, is the founder's entry on TrustMRR. | [1, 2] |
| Monthly revenue, November 2025 to September 2026 | from US$17,427 down to US$5,860 | A | Stripe data through TrustMRR: US$17,427, 13,455, 11,577, 9,602, 10,347, 8,862, 7,303, 9,943, 6,498, 6,220 and 5,860. The archived September copy gives the same values. | [1, 2] |
| Fall from the best month to September 2026 | 68% | calc | Calculated: 1 minus 5,860 / 18,298, over 11 months. Revenue was lower than in the month before in 9 of those 11 months. | [1, 2] |
| MRR and active subscriptions, 14 October 2026 | US$5,408 and 193 | A | Stripe data through TrustMRR. On 14 September 2026 they were US$6,411 and 221, so MRR fell 15.6% in 30 days. | [1, 2] |
| Total revenue, February 2025 to 14 October 2026 | US$243,652 | A | Stripe data through TrustMRR. US$118,592 of it came in the last 12 months and US$5,411 in the last 30 days. | [1] |
| Revenue from February to September 2025 | about US$115,991, or US$14,499 a month on average | calc | Calculated: the all-time total minus the 13 published months from October 2025 to 14 October 2026. These are the months before the public monthly data. | [1, 2] |
| TrustMRR figures on 15 March 2026 | US$10,507 in 30 days, 439 active subscriptions | A | Archived TrustMRR page. All-time revenue was US$194,808 and TrustMRR's estimated MRR US$13,750. | [3] |
| MRR posted by the founder, April to June 2025 | US$3,219, then US$10,000, then US$17,000 | C | Posts on X on 23 April, 20 May and 14 June 2025. On 24 June and 2 July he wrote that it makes around US$20,000 a month. | [5, 7, 8, 9] |
| Revenue in August 2026, the month of a newsletter's US$17K MRR headline | US$6,220 | A | The Startups.fyi newsletter of 11 August 2026 repeated the founder's figure from the Starter Story interview of 9 July 2025. Revenue in August 2026 was 37% of that figure. | [12, 4, 1] |
| Customers, signups and traffic, July 2025 | more than 1,000 paying customers, about 20,000 signups, about 20,000 visits a month | C | Founder in the Starter Story interview. He says the signups came 'purely from Reddit'. | [4] |
| Reach from Reddit | over 1 million views, at no cost | C | Founder in the interview and on X, 3 July 2025, when he had 115 followers on X. | [4, 10] |
| Costs, July 2025 | AI APIs about US$2,500 a month, Cursor US$20, Vercel about US$40; margin about 70% | C | Founder in the Starter Story interview. The AI APIs were the largest cost. | [4] |
| Asking price | US$27,500, listing dated 11 September 2026 | C | Seller's listing on TrustMRR. The price was US$120,000, then US$75,000 in March 2026, and US$45,000 in June 2026 according to a third-party capture. | [1, 3, 18] |
| Asking price as a multiple | 0.42 times MRR times 12; 0.23 times the last 12 months of revenue | calc | Calculated: 27,500 / (5,408 x 12) and 27,500 / 118,592. TrustMRR shows 0.4x. At the profile's 90% margin on US$5,411 a month, the price equals about 5.6 months of profit. | [1] |
| Team | 1 person, no employees | C | Founder's TrustMRR entries. He looked for a marketing co-founder on 8 May 2025 and for a part-time Next.js developer on 14 October 2025; no source shows that anyone joined. | [1, 11] |
| Outside funding | none | C | TrustMRR entry 'Bootstrapped'. The design tool's FAQ says it is not a venture-funded startup. | [1, 17] |
| Users and customers to date, founder's count | 230,000 registered users, 4,700 paying customers | C | Founder's TrustMRR text, October 2026. The same text gives a 'Peak month $19K net, Aug 2025', which the public Stripe months, starting in October 2025, cannot confirm. | [1] |
How to read this table: a grade A value comes from Stripe through TrustMRR's API connection. A grade C value is the founder's own statement in posts, an interview or his TrustMRR text, or a third party's report. A "calc" value was computed on this page from the rows above it.
| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Design tool Starter | 17.50 | 29.99 | 4 apps, 3,000 AI credits a month |
| Design tool Pro | 29.17 | 49.99 | 10 apps, 5,000 AI credits a month |
| Design tool Ultimate | 58.33 | 99.99 | unlimited apps, 10,000 AI credits a month |
| Design tool Enterprise | 199.98 | unlimited apps, 20,000 AI credits a month, priority support | |
| App builder Pro | 20 | 100 credits a month, code export | |
| App builder Max | 200 | 1,000 credits a month, priority throughput |
The design tool's yearly plans give 5 months free; the founder's TrustMRR text gives them as US$210, US$350 and US$700 a year, shown here per month. The app builder's page shows monthly prices only, and the founder's TrustMRR text gives US$100 for Max where the page shows US$200. Neither page lists a free plan, and used credits are not refunded.
The timeline
From a two-week first version to a for-sale listing at US$27,500, in 20 months. Each entry carries its grade.
- Early 2025CHe pays a designer on Upwork hundreds of dollars for an app design, and extra for each small change, and decides to build an AI tool for the job. [4]
- 11 February 2025CFounding date on TrustMRR. He builds the first version in about two weeks and starts posting on Reddit at once. [1, 4]
- 23 March 2025CHis first post on X that names AppAlchemy.ai, as a tool to 'create apps in hours'. [11]
- 23 April 2025CPosts MRR of US$3,219. [5]
- 25 April 2025CPublishes his Reddit method with screenshots of his posts' metrics: 'you can get your first 100 paid users through Reddit with 0 marketing cost'. [6]
- 8 May 2025CLooks for a marketing and distribution co-founder. No source shows that one joined. [11]
- 20 May 2025CPosts that he reached US$10,000 MRR. [7]
- 14 June 2025CPosts US$17,000 MRR 'in couple months with no audience': Reddit first, then influencers on X and Instagram. [8]
- June 2025CAdds yearly plans. On 24 June and 2 July he writes that the product makes around US$20,000 a month, and he starts building a separate consumer app. [9, 11]
- 9 July 2025CStarter Story interview: about US$17,000 MRR, more than 1,000 paying customers, about 20,000 signups from Reddit, about 70% margin. [4]
- October 2025AUS$18,298 in revenue, the best month in the public Stripe data. [2]
- 22 October 2025CPosts 'Day 1 of growing Kodo to 10K MRR', about a different product. [11]
- 15 March 2026AUS$10,507 in revenue in the last 30 days and 439 active subscriptions; US$194,808 in total. [3]
- 15 March 2026CFor sale at US$75,000, down from US$120,000. The profile says revenue fell after he 'stepped back from marketing to focus on improving the product'. [3]
- June 2026CAsking price US$45,000 and MRR of US$7,906, according to a third-party capture of the TrustMRR page dated 27 June 2026. [18]
- July 2026CLaunches the AI app builder at appalchemy.ai, which writes React Native code. By October the founder counts 6,500 signups for it. [1, 14]
- 11 September 2026CDate of the current listing at US$27,500. The seller says he is burned out, has not marketed since early 2026, and priced it to close by 30 September. [1]
- 14 October 2026AStill listed, with 97 offers received. MRR US$5,408 from 193 active subscriptions; US$243,652 in total revenue. [1]
Why it worked
Four causes, in the order they mattered. Each one is a reusable lesson.
A problem he had paid for
Before AppAlchemy he paid a designer on Upwork hundreds of dollars for an app design, and extra for each small change. He built a tool that does the same job in a chat, which he described as 'like Cursor but for designing mobile apps', and released the simplest version after about two weeks.
Reddit posts that teach first
His method, from the interview and his posts: use Reddit daily first, because posts from a new account are filtered out; list 5 to 10 related subreddits; write one useful post about the problem, with the product link in the middle; post it in all of them. Where a subreddit forbids self-promotion, also link a complementary product, in his case Cursor. He says this gave over 1 million views at no cost.
Credits tie the price to the model cost
Neither pricing page lists a free plan. Each plan includes a monthly number of AI credits, and used credits are not refunded because, in the FAQ's words, the costs of generative AI are extremely high. In July 2025 the AI APIs cost about US$2,500 a month against about US$17,000 in MRR. Yearly plans came in June 2025; today they give 5 months free.
Influencers after Reddit
After Reddit he worked with influencers on X and Instagram, on short-form video, although people told him that this would not work for a business product. His own following was small: the Starter Story host found 70 followers on his X account in mid-2025.
How it was built
These details come from the founder's interview of July 2025, his posts, his TrustMRR text and the company's own pages. They are grade C. [4, 1, 11, 13, 14]
- Next.js and React, written with the AI code editor Cursor, which cost US$20 a month in July 2025. On 25 June 2025 he moved to the Cursor Ultra plan.
- UI component libraries such as Chakra UI or Ant Design, and only widely used frameworks, because, he says, AI writes better code for the libraries that are most used on the web.
- Firebase for login and database in July 2025. The October 2026 profile lists Supabase and Firebase backends.
- Hosting on Vercel, about US$40 a month in July 2025.
- AI models through APIs, about US$2,500 a month in July 2025, the largest cost. The profile's tech stack lists OpenAI and Anthropic; no source names the exact models.
- The app builder writes React Native code that runs live on the user's own phone after a scan of a code; the user owns the code and can export it. Sign-in works with a Google account.
- Payments through Stripe, one account for both products. The profile shows a 90.0% profit margin for the last 30 days; no source shows how the costs behind it were measured.
Tools named in the sources
Each one opens its page in the Tools tab, with every story that uses it.
The dead that came before
Five companies in the Loot Drop database promised that people without design or coding skills could make apps, or sold tools to design them. The values below are from that database and are not independently verified.
Builder.ai
The promiseCustom apps for non-technical customers, built by an AI-powered assembly line.
Why it diedIn the database's words, it sold US$20,000 apps that cost US$35,000 to build, while offshore developers did work presented as AI.
Appcubator
The promiseA drag-and-drop tool to create apps with no coding.
Why it diedThe market was not ready for no-code, and more flexible tools such as Bubble won.
Crejo.Fun
The promiseA drag-and-drop mobile app builder so that creators in India could own their audience.
Why it diedCreators needed more followers, not another app to manage and promote.
Pixate
The promiseNative mobile app prototypes without writing code.
Why it diedGoogle bought it, and design work moved to web-based tools such as Figma.
Invision
The promiseOne platform to take a design from designers to developers.
Why it diedFigma's real-time collaboration in the browser took its customers.
What changed
The five predecessors either had people do the work behind an AI label, as the database says of Builder.ai, or sold a drag-and-drop editor that was too limited, or lost to Figma. AppAlchemy's language models do write the screens and the code, so one founder could serve over 1,000 paying customers for about US$2,500 a month in AI costs (July 2025), without staff or investors.
What did not change is the competition. Pixate and InVision lost to a larger platform, and AppAlchemy now competes with Figma's own prompt tool, Figma Make (May 2025), and with app builders that raised hundreds of millions of dollars. The founder says revenue fell because he stopped marketing; by his account, new customers came from his own posting, not from the product alone.
Can one person copy it now?
The Reddit method, yes. The product category, much harder in 2026.
Shipping a simple paid tool in weeks and posting useful Reddit threads in many subreddits is a method anyone can repeat for almost no money. The product is harder to repeat: AI app design and app building are now crowded with funded companies and Figma itself, and AppAlchemy's own revenue fell by two thirds within a year.
Where it is still open: Narrow tools where an AI chat replaces a complex editor. The founder's own suggestion in June 2025: 'take product you have used that has a cumbersome drag n drop/no-code UI and replace that with a chat interface and generative ai.'
- 1Build for a cost you paid yourself, and measure it: the designer's fee and the price of each change.
- 2Ship the smallest useful version in weeks and, as he advises, start selling it, in credits that cover the model cost.
- 3Use Reddit daily before you post, list 5 to 10 related subreddits, and post one useful thread in all of them with the link in the middle.
- 4Keep marketing after the first success; this founder says revenue fell once he stopped.
What to watch
The risks are the mirror image of the causes above.
Revenue fell by two thirds
Stripe shows US$18,298 in October 2025 and US$5,860 in September 2026, a fall of 68% in 11 months, with lower revenue than in the month before in 9 of them. MRR fell 15.6% in the 30 days to 14 October 2026, and active subscriptions went from 439 in March 2026 to 193. The founder's reason is that he has not marketed since early 2026; no source shows churn or cohort data.
A crowded field of funded builders
Lovable raised US$200 million at a US$1.8 billion valuation in July 2025, US$330 million at US$6.6 billion in December 2025 and US$400 million at US$13.3 billion in August 2026. Replit raised US$250 million at US$3 billion in September 2025. Bolt.new raised a US$105.5 million Series B and was valued at about US$700 million in January 2025, according to Sacra. For mobile apps, Vibecode raised a US$9.4 million seed round (reported in August 2025), Rork a US$15 million seed round (April 2026), and Figma launched Figma Make on 7 May 2025. In July 2026 the founder wrote that software had been the ideal business for solo founders because of its scalability, 'high barriers to entry' and location freedom, and that the change was felt 'a lot stronger since ~Opus 4.5'.
Apple's rules for generated apps
App Review Guideline 4.2.6 says: 'Apps created from a commercialized template or app generation service will be rejected unless they are submitted directly by the provider of the app's content.' Guideline 4.3(b) says: 'Don't submit apps that are indistinguishable from what's already widely available.' So customers must submit the apps under their own developer accounts, and each app must differ from what is already available. The app builder's pricing page offers a live preview and code export; it does not describe App Store submission.
For sale at 0.4 times revenue
The asking price is US$27,500: 0.42 times the current MRR times 12, 0.23 times the last 12 months of revenue, and about 5.6 months of profit at the profile's 90% margin. It was US$120,000, then US$75,000 in March 2026 and US$45,000 in June 2026. The seller says he is 'burned out', is 'exploring new projects', and priced it 'under what i'd keep collecting' to close by 30 September 2026. On 14 October 2026 it was still listed, with 97 offers received.
Numbers that do not match
The seller's message says 300 active subscriptions (240 monthly, 60 annual); Stripe shows 221 on 14 September 2026 and 193 on 14 October 2026. In July 2025 the founder said more than 1,000 paying customers at about US$17,000 MRR, which is under US$17 each, below the cheapest monthly plan of US$29.99; yearly plans or counting past customers could explain it, but no source says. His 'Peak month $19K net, Aug 2025' cannot be checked, because the public Stripe months start in October 2025.
Two products, one Stripe account
No source splits the revenue between the design tool and the app builder. The founder writes that the design tool 'drives most revenue' and that the builder, launched in July 2026, had 6,500 signups. Many daily totals equal sums of design-tool prices, for example US$79.98 = US$29.99 + US$49.99. The terms and privacy pages name no company, address or governing law, so the public pages do not show which legal entity owns the products.
Sources
The grade next to each source is the best grade that the source can support. "Read in full" means that the text of the page was downloaded and the numbers were checked in it.
- A TrustMRR, AppAlchemy markdown profile, read 14 October 2026Read the markdown profile in full (data as of October 2026, revenue last synced 14 October 2026). Revenue, MRR and subscriptions come from Stripe through an API key, grade A. The description, additional info, team size, funding, tech stack and the seller's message are founder-written, grade C.
- A TrustMRR, AppAlchemy markdown profile, Wayback Machine copy of 14 September 2026Read the markdown profile in full (archived copy). Source for the October 2025 month, which the live profile no longer shows; the other months match the live profile.
- A TrustMRR, AppAlchemy page, Wayback Machine copy of 15 March 2026Read the archived page text in full. Stripe-verified totals of 15 March 2026 (grade A), the asking price then, and the founder's sentence on the decline (grade C). The monthly chart loads by script and is not in the copy.
- C Starter Story, 9 July 2025: How I Used Reddit to Hit $17K MRR (With ZERO Audience)Read the page and its interview transcript in full. The transcript is machine-made and writes the product name as 'App Akamai'. Founder statements to the host, Pat Walls. Also published as a YouTube video the same day.
- C Diego Roshardt on X, 23 April 2025: MRR $3,219Read in full through api.fxtwitter.com.
- C Diego Roshardt on X, 25 April 2025: first 100 paid users through RedditRead in full through api.fxtwitter.com. The screenshots of post metrics were not read.
- C Diego Roshardt on X, 20 May 2025: $10k MRRRead in full through api.fxtwitter.com.
- C Diego Roshardt on X, 14 June 2025: $17k MRR with no audienceRead the first post and replies 2 and 3 of the thread through api.fxtwitter.com. Reply 1 was not in the timeline pages.
- C Diego Roshardt on X, 2 July 2025: makes around $20k/monthRead in full through api.fxtwitter.com.
- C Diego Roshardt on X, 3 July 2025: over 1M impressions from Reddit, then influencersRead in full through api.fxtwitter.com.
- C Diego Roshardt's X profile and timeline, posts from February 2025 to September 2026Read the profile and every timeline page through api.fxtwitter.com. Source for the location, the co-founder search, the yearly plans, the Cursor Ultra plan, the consumer app, the developer search, the Kodo post and the July 2026 post on software.
- C Startups.fyi newsletter, 11 August 2026: $17K MRR From Reddit AloneRead the section on App Alchemy. It repeats the July 2025 Starter Story figure and links to the video.
- C AppAlchemy app builder pricing page, read 14 October 2026Read in full.
- C AppAlchemy terms of service, read 14 October 2026Read in full. Says the service generates React Native code; names no company entity, address or governing law.
- C AppAlchemy privacy policy, read 14 October 2026Read in full. Sign-in via Google; no company entity named.
- C AppAlchemy design tool pricing page, read 14 October 2026Read in full. The monthly prices are in the page text; the yearly prices are not, so they come from the founder's TrustMRR text.
- C AppAlchemy design tool FAQ, read 14 October 2026Read in full. Source for HTML export, the refund rule and the statement that it is not venture-funded.
- C SoFarBot, 20 September 2026: AppAlchemy, the two-week mobile-design MVP, Reddit distribution and the truth behind $17K MRRRead in full. A third-party summary of the Starter Story transcript and of a VibeCrowd capture of the TrustMRR page. Used only for the June 2026 asking price and MRR; the capture itself was not read.
- C Apple, App Review Guidelines, read 14 October 2026Read sections 4.2 and 4.3. Apple's own published rules.
- B TechCrunch, 17 July 2025: Lovable becomes a unicorn with $200M Series A just 8 months after launchRead in full.
- B TechCrunch, 18 December 2025: Vibe-coding startup Lovable raises $330M at a $6.6B valuationRead in full.
- B TechCrunch, 12 August 2026: Lovable confirms new $13.3B valuation, raises another $400MRead in full.
- B Bloomberg News on Bloomberg Law, 10 September 2025: AI Coding Startup Replit Valued at $3 Billion With New FundingRead the opening paragraphs.
- C Sacra, Bolt.new company pageRead the funding and revenue sections. A research firm's estimates.
- B InfotechLead, 9 April 2026: Rork raises $15 mn to scale AI-powered mobile app development platformRead in full.
- B Grit Daily, 19 August 2025: Vibecode Boosts App Democratization with New $9M Seed RoundRead in full. The text gives US$9.4 million.
- C Figma blog, 7 May 2025: Introducing Figma MakeRead the header and the date. Figma's own announcement.
- C Loot Drop database export, 1,749 failed startupsLocal export. AI-written summaries, not independently verified. Source for the failed predecessors.