LinkedIn sued his $10M ARR business. Six months later, his next product hit $15k/month.

Your hardest workday problems may already have a prompt for them.

Every manager faces the same time-consuming tasks: difficult emails, hiring decisions, reports, planning, and team communication. Why start from scratch every time?

Get 37 proven Claude prompts built for real workplace situations. Copy, customize, and get hours of work done faster, plus join 400K+ readers getting a 5-minute daily AI brief.

Some setbacks arrive one at a time. Steven Goh's arrived in a stack.

Steven is a software engineer and serial entrepreneur with more than twenty years behind the keyboard and two company sales to his name. His previous business, Proxycurl, was the largest LinkedIn scraping API on the market, generating $10M ARR. In January 2025, LinkedIn sued. He settled out of court and sold Proxycurl to a competitor.

Shortly after the acquisition closed, he fractured the metatarsal bone in his right foot. Immobile, with time on his hands and Claude Code open, he prototyped a new product in two weeks and launched it: NinjaPear, a B2B company intelligence and data enrichment platform.

That was six months ago. Last month, NinjaPear did over $15k in gross revenue. Steven runs it solo, full time, and it is already profitable.

Here is how he picked the idea, why the lawsuit shaped the product's entire architecture, and the failure philosophy underneath all of it.

The idea for NinjaPear came directly from the wound. Steven knew the B2B enrichment data market as well as anyone, having built the category's largest player. What he wanted this time was the same business without the same vulnerability: a Proxycurl built on his own data instead of LinkedIn's, capable of scaling for the next ten years without what he calls the Microsoft legal war machine hanging over it.

The technical approach makes that possible. NinjaPear uses LLMs performing deep web research in real time to assemble rich B2B data - none of it sourced from LinkedIn. That independence also produces a product advantage: NinjaPear surfaces data points LinkedIn simply does not have, including customer list data, competitor research, and product lists.

The build cost was two weeks of time and a Claude 20x Max subscription. Steven is candid that money is not a constraint for him - he was an early cryptocurrency investor and has two exits behind him - which means he can work on NinjaPear indefinitely without profit pressure. It happens to be profitable anyway.

Steven's view on technology choices in the post-AI era is blunt: the tech does not matter. His stack is boring on purpose:

The one genuinely unusual choice is physical: he owns all the servers and co-locates them at his house, running them on solar power. For a solo founder operating a data platform, owning the hardware outright removes a whole category of recurring cost - and the electricity bill along with it.

NinjaPear's model is credit-based. Users pay for data with credits, purchased either through subscriptions (cheaper) or pay-as-you-go top-ups. The business depends on customers integrating NinjaPear data into their workflows - when their usage grows, revenue grows with it.

The biggest challenge right now is churn, and Steven has specific hypotheses about the causes. The first is speed: NinjaPear's data processing gets compared to LinkedIn's, and it is slower, because LLMs performing web-scale research in real time take longer than serving pre-scraped records. The second is product messaging, which he acknowledges is not yet landing the way it needs to. He is iterating on both constantly.

What is notable is his relationship to the problem. He does not frame churn as a crisis. He frames challenges as the substance of the work itself - the reason venture building is worth doing at all.

Steven's marketing effort is deliberately narrow. Three channels:

He plans to keep wearing the go-to-market hat personally for the next six months. The domain retention detail is worth pausing on: when selling a business, most founders let everything go. Keeping the domain was a quiet, structural bet on his own next act.

The most distinctive part of Steven's operating philosophy is his deliberate relationship with failure. He takes nothing personally anymore - not failures, not critics. His working mantra is to set out to fail ten times, then keep failing and recalibrating as he goes. What matters, in his framing, is the cadence of output. Keep going. Keep failing.

He gave a concrete recent example. A few days ago, he spent a couple of hours writing a detailed article on Reddit about building a $1M ARR venture with AI as a solo founder. It got barely any traction. A younger version of himself would have been genuinely upset. The current version reads it as a signal to try again - and again - until something eventually travels.

Try again tomorrow. That is the whole method.

Steven's brother has attempted entrepreneurship a couple of times and remains an employee. The advice Steven gives him distills twenty years into two points:

The goal from here is $1M ARR as a solo founder - a target Steven regards as entirely achievable given where NinjaPear already is six months in. Between the profitable base, the retained SEO foundation, the 20k-subscriber list, and a founder with no financial pressure and no intention of stopping, the pieces are in place. The open questions are the ones he is already working: churn, processing speed, and messaging.

You can follow the build on X, subscribe to his 1MO Solo Founder Newsletter, or look at the product itself at NinjaPear.

Subscribe to The AI Report, the free 5-minute daily AI brief for 400,000+ business leaders, and you’ll get 37 Claude prompts free in your welcome email. They’re organised by the 8 situations every manager faces. You get both: the newsletter and the prompts.

Subscribe & Get the Prompts