Built on another company's platform
These products live inside or on top of another company's product: a browser extension, an add-in, an app in a store, or a tool for the users of one platform. The host brings the users. The host can also change its rules, its prices or its own features, and the product must follow.
Built on another company's platform holds 22 of the 110 stories on Aiolum: 7 run by a solo founder and 15 by a team of 2 to 10. Together they report US$10M in verified outcomes; the largest is Tweet Hunter, Sold for US$2M cash plus earn-out, August 2022. 14 headline numbers carry evidence grade A (disclosed or verified), 0 grade B (independent press) and 8 grade C (founder claim). The median story reached its headline number 17 months after launch. The founders are in 14 countries, most often USA and France. The most common main channel is inside another platform (6 of 22). The tools named most often in their sources are Stripe (18), Next.js (14) and TrustMRR (14).
One block per story.
Darker earth means stronger evidence.
Verified, number by number
Browse by what the product does
Patterns to copy
More filters: main channel, size and time
The revenue trend compares the latest months in the sources with the months before them. Falling means revenue is about 15% or more below the best recent months, or the sources report a fall. The headline number is the yearly figure or the sale value on each card.
22 of 110 stories shown
A user gives a topic, a text, a PDF, a web address or a YouTube link, and a language model writes a slide presentation from it. MagicSlides runs as an add-on inside Google Slides and as a web application that exports editable PowerPoint, PDF and Google Slides files. Since July 2026 the same account also includes AI tools for spreadsheets, documents, meeting notes and a design canvas, under the name Magic Suite.
It appeared in January 2023 inside Google Slides, a program where people already work, when writing slides from one sentence was new, and users showed it to each other without paid advertising. A blog about presentation topics then brought about 80% of new users from Google search, by the founder's account. Both channels weakened: search traffic fell, and Google added the same function to Slides. The verified total is real, and so is the fall: monthly revenue fell by half between February and July 2026.
An AI agent for paid advertising that works inside ChatGPT, Claude, Claude Code, Cursor and other AI assistants. It connects to the customer's ad accounts at Google, Meta, LinkedIn, TikTok, Amazon and ChatGPT Ads, researches keywords, builds campaigns (always created paused), reports results and finds wasted spend. Scheduled agents can also watch the accounts and make reversible changes. The customer pays the ad platforms directly.
It went where the users already were: an app in ChatGPT's app store, a plugin in Claude's directory and in Claude Code, and setups for Cursor and Codex, in the months when these directories were new. It charges per tool call instead of a share of ad spend, so a small advertiser starts at US$49 a month. And it applied to the platforms' own API programs; the company says Google approved it in June 2026 and Amazon admitted it to its partner network.
A LinkedIn lead finder for B2B sales teams. You list LinkedIn profiles to watch; traxy collects the people who like, comment on or repost their posts, scores each one against your ideal buyer, adds an email address and phone number, and an AI agent can write and send the first message.
The founder already ran a LinkedIn ghostwriting agency and built the lead tool that the agency needed. He sold it with the tool itself: traxy found the leads, and his team contacted them. Daily posts on X added trials, and credit-based prices let heavy users pay more.
A web application for people who sell simple books through Amazon's Kindle Direct Publishing. The user describes a book, and AI models produce its pages: line drawings for coloring books, puzzles and mazes for activity books, text and layouts for journals and planners, and a description for the Amazon listing. A subscription buys credits. By the terms of service one generated image costs 200,000 credits, so the smallest plan, with 40 million credits a month, pays for about 200 images.
It sold one narrow job to a group that already existed: people who learn from YouTube channels and paid courses how to publish many simple books on Amazon. The product carries that group's own word in its name, tutorial channels in English and Italian showed it from January 2025, and the site runs an affiliate script. Two thirds of the US$642,409 came in the first 14 months, at about US$30,300 a month. The last 12 months average about US$17,600 a month.
A web app that writes LinkedIn posts in the member's own style from an idea, a voice note, a video, an article or a short interview with an AI, then schedules and publishes them. Team plans add approvals, a shared calendar and analytics for each employee. It can also be used from inside Claude and ChatGPT.
It entered a market where people already paid for tools such as Taplio, and did the one core task better: writing the post. The founders' public posts brought the first customers and a lifetime-deal platform, and that deal paid for growth without investors. Subscriptions, partnerships with LinkedIn influencers, affiliates and a move to team plans came after.
A desktop app and Roblox Studio plugin that builds parts of a Roblox game from a written request: game systems, user interfaces, animations, visual effects, sounds and 3D builds, written straight into the user's project after the AI has read it. Users pay for AI tokens with a monthly plan or one-time top-ups, and can install ready-made systems from a marketplace.
The founder had made Roblox games for nine years, with 200 million visits on his own games, and his launch post on the official Roblox developer forum was read 112,656 times. He launched free, with 1 million tokens a month, and charged US$20 a month for heavy use. Then he added top-ups and a marketplace of tested systems that the AI installs, instead of writing every system from nothing.
An AI agent for Upwork. It reads the Upwork job feed against filters that the customer sets, checks with AI whether each job fits, writes a proposal and the answers to the screening questions from the customer's own knowledge base, and submits it from the customer's Upwork account, with optional boosting. Replies are tracked in an inbox. An MCP server and a REST API let the customer's own AI agent run it.
The founder had the problem himself. Upwork had become his agency's best sales channel, but 10,000 jobs a day are posted around the clock and he could cover only about 8% of them by hand. He built the tool for his own agency and charged agency owners a business price from the start. Two Upwork coaches, who advise freelancers and agencies, switched their clients to it as affiliates, and that produced most of the growth.
A dictation app for Mac, Windows, Linux, Chrome, iPhone and Android. You press a keyboard shortcut and speak, and the text appears with punctuation in whatever app or website you are typing in. The speech is transcribed in the cloud by the open-source Whisper model through Groq's API, or, since 12 October 2026, offline on the computer.
It started as a free Chrome extension that the founder built for his father, then sold unlimited dictation everywhere for a low monthly price. The AI model is cheap to rent: Groq lists Whisper at US$0.04 to US$0.111 per hour of audio. Users came from Google: the site has more than 1,000 pages, one for each app and each need, and the founder says that an AI agent he made writes the blog posts.
An add-in for Microsoft Word, plus a web app, for German-speaking students who write seminar papers and theses. The student uploads the PDFs of the literature. While the student writes, fastwrite suggests the next sentence or a whole paragraph with a citation from those PDFs, and the Tab key accepts it. It also answers questions across all uploaded PDFs with the exact page, searches more than 20 million papers, and checks a draft against nine criteria that universities use for grading.
It puts the AI where German students already write, inside Word and in German, and each suggestion carries a citation from the student's own PDFs, so the student does not have to fear invented sources. A free plan without a time limit brings students in, and a monthly plan sold for the months in which they write turns some of them into paying customers.
An AI assistant that you reach by message. You send a voice note, a text, a screenshot or a forwarded email on WhatsApp, Telegram, iMessage, Slack or email, and it writes notes, tasks, draft posts and CRM updates into Notion and more than 1,000 other connected apps. It started as a voice-to-Notion tool. In 2026 it added a desktop app, a cloud computer, automations, and connections to Claude, ChatGPT and coding agents.
It reached people inside the chat app that they already open all day, so capturing an idea on a phone took one voice note. It aimed first at Notion users, whom the company told that ideas get lost on the phone. Ads on Facebook and Google then paid for fast growth in early 2026, until the founder stopped them in August 2026 because the costs were larger than the revenue.
A remote MCP server that gives an AI assistant such as Claude or ChatGPT tools to generate images, video, speech, music and 3D models with models from Google, OpenAI, ElevenLabs and others. The user installs one plugin or adds one address, logs in once and buys credits from CreativeClaw: US$10 for 1,000 credits, with a published credit cost for each model.
People already pay for an assistant and already ask it for pictures and video. CreativeClaw sells inside that conversation, so it needs no interface of its own, no sales calls and no provider accounts on the customer's side. Revenue stayed under US$5,300 a month for five months, then rose about 18 times in September 2026, after the founder had packaged the server for ChatGPT and prepared its entry in the plugin directories. He has not published the cause of the rise.
A Chrome extension that adds folders, full-text search, a prompt library, bookmarks and export in four formats to the web pages of ChatGPT, Gemini, Claude and Grok. It sells no AI: users keep their own accounts, and the search index of their chats stays in their browser.
Heavy users of AI chat apps lose old answers, because the apps had no folders, no filtered search and no bulk export. AI Toolbox adds these inside the apps, for four apps at once, sells them mostly as one-time licenses, and, by the seller's account, finds its buyers through search pages instead of ads.
A business connects its lead forms, for example Facebook lead ads, LinkedIn forms or a website form, through Zapier or Make. About 10 seconds after each form arrives, Setter AI sends the person a WhatsApp message, answers questions, asks qualifying questions, follows up when there is no reply, and books the meeting in Calendly or Google Calendar. A chat window does the same on the business's website, and WhatsApp reminders reduce missed meetings.
It sold one promise that a customer can measure: a first reply to every new lead within 10 seconds, to businesses that already pay for many leads through Meta ads. A demo on the website brought sales calls through organic search before the product was built. Revenue then fell from US$9,000 to US$12,000 a month in late 2025 to under US$2,000 in September 2026, in the same months in which the founder launched other products.
A paid library of more than 400 Agent Skills: packages of instructions, templates and scripts that an AI coding agent such as Claude Code, Codex or Cursor installs and follows to make a web page section, a slide deck, a set of images or a short video in one designed style. Every Skill page shows real output first. Some Skills are free; the others cost US$4.99 each or come with Premium at US$9.99 a month, US$79 a year or US$169 once.
AI coding agents can build a working page or deck, but in the founder's words their visual output often looks generic. Skillry sells finished visual styles, one task per Skill, shows real output before the sale, and adds new Skills almost every day under one subscription. When sales began on 28 September 2026, they held at about US$430 a day for 17 days.
A tool for people who want to grow an audience on X (then Twitter). It started as a search box over a library of viral tweets, then added a scheduler, automatic direct messages, a contact list, and a language model that proposes tweets and threads in the user's own style. A sister product, Taplio, does the same for LinkedIn.
Two founders shipped eleven products in about four months and kept the one that made money. They sold it on the platform it served, gave an audience owner a share of profits instead of a sponsorship fee, and used a language model to make writing faster, not to replace it. Fifteen months after launch, with about US$1 million a year in revenue, they sold to a larger French software company run by a former schoolmate.
A web app for LinkedIn posts. It stores what a member knows and how the member writes, suggests topics from posts that do well, writes drafts with the member in a chat, makes 20 images a month, and schedules and publishes posts to a connected LinkedIn account, with statistics and weekly group calls. A Chrome extension saves ideas from other websites.
Three of the four co-founders had LinkedIn audiences, together more than 480,000 followers, made of the people Kleo sells to, and the free extension had already reached tens of thousands of users. When LinkedIn stopped the extension, they sold a rebuilt paid version to that audience first, in limited groups of seats, before any public launch.
An app for Shopify stores that sends marketing messages on WhatsApp and answers shoppers with an AI assistant trained on the store's catalog: reminders for abandoned carts, campaigns to subscribers, product advice in 96 languages. The store pays a monthly plan plus a charge for the WhatsApp conversations.
After a first product of their own that nobody bought, the founders copied a kind of Shopify app that already earned money, and sold it in France, where few competitors sold it. They found customers only by cold outreach, with tools that cost about US$380 a month. By their account they sold the company seven months after the App Store launch, when sales outside France became hard.
A web app and a Chrome extension that write LinkedIn posts. The member picks an industry, a niche and a type of topic, or gives an idea or a voice note, and the AI writes a draft in the member's style; the member edits it and posts it. By 2026 it also showed the best posts of top creators, found trending topics on Reddit, X, Google, Bing and Perplexity, and planned posts in a calendar. It never posted on its own.
The founder had already built one of the largest audiences on LinkedIn by posting about AI every day, so the product had buyers on the day it launched. It sold one plan at about US$60 a month, as a US$2-a-day ghostwriter against US$2,000 to US$3,000 for a human one, and his employees grew their own accounts with it and sold it for a commission.
A company connects My AskAI to the help desk it already uses, such as Intercom or Zendesk, and gives it the help center, the website and past tickets to read. The agent then answers customer questions inside that help desk, or writes a draft as an internal note for a human agent. When it cannot answer, it hands the conversation to a person in the same tool, with a summary.
The founders sold the idea before they built it, and built it with a no-code tool so that one person could change it every day. Then they cut the product down to the one use that brought 70% of the revenue. They placed it inside the help desks that customers already pay for, and priced it below the AI agents that the help-desk vendors sell themselves.
A web app, a Chrome extension and an MCP server for Amazon sellers. It estimates the revenue, prices, reviews and keyword demand of a product market, matches Amazon products to suppliers on Alibaba, analyzes ads, tracks rankings and drafts listings. Paid plans connect the seller's own Amazon account, so Claude or Codex can answer questions about the seller's sales and ads.
An Amazon seller who could not code built the research tool he needed in 48 hours with Cursor, then offered it to Legacy X, a coaching program for Amazon sellers that he had bought two years earlier. Legacy X took a share of the business, and its students were the customers from the first day. He says it reached US$10,000 a month in 30 days and US$30,000 a month by June 2026.
A free browser extension for the ChatGPT website that adds folders and subfolders, search in the full chat history, bulk export to TXT, Markdown, JSON and PDF, pinned messages, a prompt library and a tree map of conversation branches. A Pro plan removes the limits. It works inside the user's own ChatGPT account.
It shipped within a week of ChatGPT's release, when ChatGPT had no folders and weak search, and it was free, so it spread through the Chrome Web Store. A daily AI newsletter shown inside the extension turned the users into an audience that sponsors paid to reach, and a Pro plan added subscriptions ten months after launch.
An AI writer for an online shop's automated marketing emails. Connected to Klaviyo or another email platform, it writes a different subject line, headline and text for every recipient from the products that person looked at or left in the cart, in the shop's tone, in more than 40 languages, and measures the result against the shop's old email.
Online shops send the same abandoned-cart and viewed-product email to every customer, because writing more versions takes people. A rented language model made one unique email per recipient cheap. Alvas sold this first through a Danish email marketing agency that set it up for its clients, and showed the gains with tests against the shops' old emails.
The next verified one-person AI business goes here. Same structure, same evidence grades.
Every number on these cards carries an evidence grade. How Aiolum verifies explains the four grades and the five parts of each story.