Back to the newsroom

Interview

How Korea Deep Learning chose document AI over trends and was profitable from the year after its founding

Korea Deep LearningPublished

How Korea Deep Learning chose document AI over trends and was profitable from the year after its founding
In an October 2025 interview with Byline Network, Korea Deep Learning said it has grown profitably without outside investment since the year after its founding.

Korea Deep Learning is a document AI company that has been profitable since the year after its founding and has grown on its own revenue alone, without outside investment. On October 16, 2025, Byline Network met Kim Ji-hyun, CEO of Korea Deep Learning, for Basri (Byline Startup Review), its startup interview column, and looked closely at how the company has grown. The column features founders whom its reporters have met in person.

Where would a business owner spend money?

Before its founding, Korea Deep Learning started by asking which AI businesses would actually pay for. The answer was documents.

"If I were the head of a company, I would spend money on deep learning that reads and processes documents."

— Kim Ji-hyun, CEO, Korea Deep Learning

Instead of chasing trends, Korea Deep Learning focused on the clear demand for document processing and built AI that earns money in the public and financial sectors. That is where the company began.

That judgment was borne out in revenue from the very first year. In 2019, its first year, Korea Deep Learning built a deep learning waybill recognition system for Korea Post and developed an ID card recognition module for Shinsegae I&C. In 2020, it supplied business card recognition optical character recognition (OCR) to Shinhan Bank, securing its first reference in the financial sector.

Korea Deep Learning expected that, however far large language models (LLMs) advance, they still have to be trained on good data, so LLM companies too would end up using its technology for turning documents into data. The same logic holds in the field. In August 2025, in the Gyeonggi Provincial Government's generative AI platform project worth KRW 13.1 billion in total, Korea Deep Learning took on the role of using DEEP Parser to turn unstructured documents such as HWP and PDF files into structured data that AI can read.

Profitable from the year after its founding

Korea Deep Learning opened in 2019 and has been profitable since the following year, 2020. The article reported that Korea Deep Learning recorded annual revenue of KRW 3 billion and operating profit of KRW 300 million in 2024. In December 2024, five years after its founding, its cumulative revenue passed KRW 10 billion, and in March 2025 it launched DEEP Agent, its document AI agent, shifting to product-led growth.

The interview also covered the period when the company was raising its Series A. On October 1, 2025, about two weeks before the interview ran, Korea Deep Learning said it was raising a Series A in the KRW 10 billion range, led by Translink Investment. It was the first venture capital (VC) investment the company had received in the five years since its founding.

The full interview was published by Byline Network.

About Korea Deep Learning

Korea Deep Learning (KDL), founded in 2019, is an enterprise AI agent company. Built on KDL Frontier, its own vision-language model (VLM) specialized for documents, KDL makes AI agents that take on and complete the document work of businesses and public institutions. Its main products are DEEP OCR, DEEP Parser, DEEP Agent and DEEP Ops.

Source