Stan's 14-day AI build turned customer knowledge into $3 million ARR

Why it matters

Stanley shows how AI coding shifts the bottleneck from engineering speed to founder judgment: who knows the customer, who owns distribution, and who can keep output useful after the demo works.

John Hu (@JayHoovy) says Stan built Stanley, its AI content product, in 14 days and grew the product line to $3 million in annual recurring revenue, according to a Business Insider essay published July 8.

The timing matters. Stanley is not a new launch this week. Hu said the LinkedIn version launched in June 2025, reached $200,000 in ARR within six weeks, and crossed $1 million ARR in October 2025. The Instagram version followed in March 2026. The July 8 essay is Hu's retrospective on how Stan got there, and it reads as a founder's argument for a specific kind of AI-era product building: use AI to compress engineering time, then spend the saved time on customer work, distribution and fast validation.

Hu's own path explains why that argument lands differently from a generic "AI lets anyone code" pitch. He told Business Insider he worked as an investment banking analyst at Goldman Sachs in New York from 2016 to 2017, then at another investment bank in San Francisco from 2018 to 2020. He started an MBA at Stanford in 2020 and dropped out to build Stan, whose core product, Stan Store, helps creators sell digital products and courses.

Stanley is a distribution product inside a monetization business

Stan's original business is Stan Store, the digital storefront and creator-commerce product. Stanley moves up the funnel. Instead of helping creators collect payment after they have demand, Stanley tries to help them generate the audience and content that create that demand in the first place.

Hu told Business Insider that Stanley helps users develop and scale content across LinkedIn and Instagram.

That shift is useful context for the ARR breakdown Hu provided. He said Stan is at nearly $41 million in total ARR: $38 million from Stan Store and $3 million from both versions of Stanley. Those figures are self-reported by Hu.

The 14-day build started before the product existed

The strongest part of Hu's account is that vibe coding came after the customer problem had already been narrowed.

Hu told Business Insider that he personally acted as "the Stanley behind the curtain" in early interviews. He came up with content ideas for prospective customers, drafted sample emails and asked whether they would subscribe. That is old-fashioned concierge MVP work dressed in AI-era language. Before Stan committed engineering time, Hu was testing whether creators cared about the output.

That matters because AI coding tools can make bad product judgment more expensive, faster. Hu's phrase for the failure mode is "coding slop." His point is plain: AI can help build almost anything, so the founder's job shifts toward deciding what deserves to exist. In Stanley's case, Stan already had a base of creators using Stan Store and a founding team that had built audiences itself. Hu said that customer familiarity made the fast build possible.

The go-to-market motion was equally direct. Hu said Stan reviewed about 200 LinkedIn creator profiles to find early beta customers. Stanley then wrote cold emails with content ideas tailored to those creators. Stan added a link inviting recipients to draft posts or keep using the product. The click became a demand signal before Stan had to pretend the market was larger than it was.

Hu also said public posting helped create the first wave of users. Business Insider's essay contains two related numbers: Stanley for LinkedIn hit $200,000 ARR within six weeks, and Hu said Stan reached its first $200,000 ARR after two weeks of posting publicly. The clean read is that public building helped accelerate early demand inside the broader six-week post-launch window. The narrower truth is still important: Stan used Hu's audience as a distribution channel before spending heavily to create one elsewhere.

The open question is quality at scale

Stanley has early revenue by Hu's account. The harder test is whether an AI content agent can keep producing output that feels useful once it moves beyond handpicked beta users and founder-led onboarding.

Hu says the team were their own early customers and content creators, which gives Stan real internal users, but it also means the initial proof comes from people already unusually motivated to publish online.

The next customer is harder: the creator or founder who wants distribution, lacks a clear voice, and does not already have Hu's habit of building in public. Stanley has to make that user better without flooding LinkedIn and Instagram with interchangeable AI posts. Hu's "coding slop" warning applies to content too.

Business Insider's essay does not disclose Stanley's pricing, retention, churn, customer count, profitability, funding history or valuation. Hu's ARR figures make Stanley a meaningful product line inside Stan, though still a small one compared with Stan Store. The strategic read is that Stan is using AI to protect the storefront business by moving closer to the creator's daily workflow. The founder lesson is narrower and more durable: the 14-day build was only possible because Hu and Stan had already spent years inside the customer problem.