Vanna AI: Product, team and how it was built

Team of 2 to 10 · 45 months after launch, revenue falling

Said US$19K a monthRecords US$9.6K in JulyDid not hold

It appeared in January 2023 inside Google Slides, a program where people already work. At that time 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.

Solo founder · 29 months after launch, revenue falling

It is run by a search marketer. Its pages target what people type into Google. Its best month came from a list article on another company's blog that ranked there. A US$1 trial that needs a card turns visitors directly into subscribers at US$25 a month or US$99 a year.

Solo founder · 36 months after launch, revenue flat

The first customers came from a lifetime-deal sale on AppSumo, at a time when the product had no AI at all. That gave about a thousand users, word of mouth and a Powered by Guidejar link on every free guide. From April 2024 the founder sold subscriptions instead, and added AI writing, voiceover and translation to the paid plans, metered in credits.

Team of 2 to 10 · 24 months after launch, revenue falling

It sold a simple promise, a finished PowerPoint file from one sentence, at a low price. In 2024 and the first half of 2025 the archived pricing showed unlimited decks for US$9.99 a month. Stripe payments read by TrustMRR add up to US$60,120 since the start. The verified monthly figures also show a steady fall after November 2025. By then the plans had moved from unlimited use to slide credits. The founder put the business up for sale.

Team of 2 to 10 · 47 months after launch, revenue falling

It put a language model inside Google Sheets from November 2022, and Google's marketplace brought more than 201,000 installs. Few installs paid: US$52,136 of verified revenue in 47 months is about US$0.26 per install. Its most important result may be indirect. A request for a pitch deck for SheetAI led the founder to build MagicSlides. By its own verified data, MagicSlides earned about 14 times as much. SheetAI's own revenue is now falling, and Google has offered a similar AI function inside Sheets since June 2025.

Solo founder · 15 months after launch, revenue falling

It does the work in the cells instead of explaining what to paste. It works in the Excel that people already have, without a Microsoft Copilot license. The founder already taught Excel to a large French audience. The AI models are bought by the request from OpenAI and Anthropic, so one person could build and run it.

Solo founder · 35 months after launch, revenue falling

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. The founder stopped them in August 2026 because the costs were larger than the revenue.

Team of 2 to 10 · 42 months after launch, revenue falling

The plugin lives inside Obsidian's community plugin store. The founder says that the store gave it organic growth with no ads or paid creators. The downloads reached 300,000 by the end of 2024, 1 million by February 2026 and 2 million by October 2026. The code is free, so there was nothing to sell until the company added a managed layer on top. The layer is hosted models, web search and tools that work without any setup. One-time lifetime plans, first Believer and then Supporter, brought large payments in single months. TrustMRR, reading the company's Stripe account, shows US$913,644 of revenue in total. Monthly recurring revenue was US$17,899 from 1,356 subscriptions on 27 October 2026, after a fall of about 51% from the best month, February 2026.

Solo founder · 14 months after launch, revenue falling

Said US$15.4K MRRRecords US$7.8K MRRDid not hold

In n8n each node has its own settings, and Synta says that "Most AI tools hallucinate n8n configs", which means they invent settings that do not exist. Synta sells the missing knowledge. It is a database of more than 800 node schemas, workflow patterns and templates, plus a loop that runs the workflow and repairs it. It is a narrow product for the users of one platform, sold as a subscription from US$34 a month.

Solo founder · 35 months after launch, revenue falling

The founder wrote that the new GPT-4 Vision and Turbo API previews allowed only 100 requests a day. Nine days later he released it as a free open-source tool. Hacker News and GitHub trending brought the repository more than 35,000 stars in under three months. The README sends readers on to the paid hosted version for people who do not want to set up keys and servers. It had made US$1,000 within 19 days, with a Stripe payment link and no user accounts.

Started small, grew after funding · 26 months after launch, revenue growing

Three founders treated each app as an experiment in pricing and paywalls. Lessons from Pixi, the first app, went into Cue. The studio rents the models and does not train them. It spends its effort on design, data and paid user acquisition. Its founders call these the three Ds.

Solo founder · 16 months after launch, revenue flat

It reached the App Store early, in July 2023, with the simplest form of the idea: press one button, get a transcript and a summary. The founder paid for Meta ads while they still worked for a new kind of app, tested paywalls constantly and moved buyers to annual plans. He works alone with AI coding tools, so in his own 2026 account about US$3 million of about US$7 million a year stays as profit.

Solo founder · 27 months after launch, revenue growing

Said US$245K a monthRecords US$263K MRRSaid less than the records

The founder took a twenty-year-old product category with no open-source entry and open-sourced all of it. This gave him millions of Docker downloads and 37,000 GitHub stars without a marketing budget. Revenue stayed small for a year. It multiplied when he turned the product toward people who automate their posting with n8n workflows and, from late 2025, with AI agents. Agents need an API-first scheduler and keep paying because they post without human effort.

Team of 2 to 10 · 8 months after launch, revenue growing

It entered a crowded category in March 2025 with paid advertising instead of a personal audience. It treated the economics of each ad channel as the product to optimize. Cheaper attribution let it test low-cost traffic. Weekly advances on App Store payouts let it put each week's earnings back into ads within days instead of 45. Adapty, the company whose software runs the app's subscriptions, published the figures under its own name.

Team of 2 to 10 · 16 months after launch, revenue growing

Every team that builds with AI models ends up with several provider accounts, keys, invoices and SDKs. One gateway with open code replaces them, and a developer can read the code before sending traffic through the hosted version. In 2026 the founders added DevPass, a flat monthly price for coding agents. Its allowance is up to three times the price in model usage. The verified monthly receipts grew from US$57,994 in January 2026 to US$249,922 in August 2026; TrustMRR does not split them by product.

Team of 2 to 10 · 2 months after launch, revenue growing

xAI released Grok Bot, a messaging app of AI agents, in beta on 11 August 2026. The two founders published their own open version about two hours later, under the name OpenGrokBot. They pitched it to people who wanted the same idea on Claude or Codex instead of Grok. Payment records show the first payment on 27 September 2026 and US$5,503 in the 30 days to 25 October 2026.

Solo founder · 19 months after launch, revenue flat

Each image or video model comes from a different company with its own API, account, price list and outages. A developer who wants to try several models must write and maintain code for each one. The same is true for an app that must keep working when one provider is down. ImageRouter puts them behind one key and one request format. It routes each call to a cheaper or working provider, and charges a margin on the credits.