Profile
The man who decided not to trust Google
Luke Jinu Kim was five years old the first time he touched a computer. By 1999 he was living in the United States, watching the dot-com boom ignite and extinguish itself in real time. That experience lodged somewhere inside him. When he enrolled at Yonsei University to study computer science, he wasn't planning on a career. He was planning on a company.
He never worked for anyone else. CEO was his first - and only - job title. He got there the expensive way: four products before Liner, a legal entity registered in 2012 while he was still an undergraduate, and a 2013 stint in South Korea's SW Maestro program that he credits with shaping how he thinks about software.
In 2015, Kim arrived in Silicon Valley with $50,000 in seed money and a bag. He rented an Airbnb in San Jose and spent months building a web highlighting tool designed to do one specific thing: help people mark the parts of the internet that actually mattered. When he flew back to Seoul, he had $300 left.
"I was struck by how challenging it was to navigate and trust the abundance of information online."
- Luke Kim, Founder & CEO, Liner
That web highlighter became a recommendation feed. The recommendation feed became an AI-powered research platform. And then, in November 2022, ChatGPT arrived. Kim moved faster than most. Within one week of the ChatGPT API dropping in March 2023, Liner had integrated LLM capabilities. Three months after ChatGPT launched publicly, Liner had an AI search product in the market - one built on a principle that was almost contrarian for the moment: show your sources. Every sentence. Every time.
"The metric we value above all else is accuracy," Kim has said more than once. In a category full of confident, occasionally wrong AI assistants, that bet is looking prescient. Liner now scores 93.7% on OpenAI's SimpleQA benchmark. It's the only Korean AI startup to appear on Andreessen Horowitz's Top 100 Gen AI Consumer Apps list - four consecutive times.
The go-to-market reads like a case study in patience. Rather than chasing enterprise contracts or consumer splash, Liner burrowed into university campuses. The logic was clean: students doing real research would instantly know whether an AI search result was accurate or hallucinated. Liner's citation-first architecture was built for that scrutiny. It worked. Today, 10% of UC Berkeley's student body uses Liner through their official school email. Similar numbers hold at USC, University of Michigan, and Texas A&M.
"We want to be the last mover, not the first mover."
- Luke Kim
The company is headquartered at 330 Townsend Street in San Francisco - the same building zip code as dozens of other high-growth startups. But the math is different inside. A 40-person team drives 30 million monthly web visitors. At the time of the a16z recognition, Kim was running an operation described by some observers as the world's most efficient AI startup. Revenue for 2024 was up 341.3% year-over-year. The $29 million Series B, closed in October 2024, brought Samsung Venture Investment into the cap table alongside Atinum Investment and InterVest.
Ninety percent of Liner's users live outside Korea. Sixty percent of paying subscribers are in the United States. Kim set up Liner's San Francisco office specifically so his team could meet users face-to-face, not because he read a playbook about customer discovery but because he once learned the hard way what happens when you stop doing it. During COVID, the team shifted toward metrics and away from direct user interviews. He says it took 18 months to correct that mistake.
He built an internal AI tool called L2-D2 that analyzes patterns in customer interview data. The name is whimsical. The purpose is structural - customer obsession baked into the operational layer, not just the mission statement.
