The case at a glance
- SocialKit (social scraping API): $3,300 MRR, 18,500+ signups
- PostPeer (social posting API): $1,700 MRR, 3 months old
- Combined: ~$5,000 MRR, zero paid acquisition that stuck
- Prior exits: $15K + $7K on two earlier -Kit products
- Products shipped: 6 total (one flop, two exits, three running)
- Blog cadence: 1-2 posts per week from day zero
- Working ad spend: $0 (Google and Meta both failed to convert)
Stack is deliberately boring: Next.js + Fastify on Railway, MongoDB, Redis, BullMQ. Infra runs under $40 a month. The operator is a solo builder on his 6th product, publishing under a .dev domain family that reads as one company to both humans and language models.
What he is actually selling in the LLM era
The product is not the API. The product is roughly fifty landing pages that ChatGPT and Perplexity quote when a user asks for a tool.
Here is the mechanism most builders miss. An API is one line of syntax a model can articulate cleanly: GET /transcripts?url=.... A dashboard SaaS is a 40-button admin panel a model cannot describe in a sentence. When someone asks a chatbot "how do I scrape TikTok comments programmatically", the model reaches for whatever it can phrase in one line. API-shaped products win that phrasing lottery every time.
The 18,500 free signups are not a conversion funnel. They are a link corpus. Every developer who tries the tool pastes it into a README, a Stack Overflow answer, a tutorial. Those mentions get scraped into search indexes and, more importantly, into next year's LLM training runs. The free tier is not leaky revenue. It is buying training data at the cost of a Railway container.
The "X vs Competitor" pages he ships alongside each feature LP intercept demand aimed at incumbents' brand names. Lateral capture: appearing inside someone else's search intent for the price of a markdown file.
He is not ranking on Google. He is training the model that will replace it.