Celon Raises Seed Funding to Build AI Memory Layer That Sharpens Response Quality

Celon, a Korean startup developing a memory layer designed to improve the response quality of AI services, has raised a seed round from Primer and The Ventures.

The company was founded and is led by Junha Jang, who is 18 years old. Over the past three years, Jang has been iterating on product development and entrepreneurship, focusing on identifying real problems users face and rapidly validating and refining solutions to address them.

Celon’s core product is Memory (Memory.inc), a platform that connects fragmented data — including conversations, internal policies, code, and emails scattered across multiple chat tools and services — to enhance the quality of responses and decisions made by existing AI services. Rather than functioning as simple storage, Memory is built so that retrieved materials, answers, and the reasoning behind judgments accumulate over time, enabling AI to operate with increasingly refined context and deliver progressively better outputs. The platform is focused on rapidly surfacing and linking both internal team data and high-quality external sources, while continuously building up information generated through day-to-day workflows into a form that AI can actively leverage.

“The quality of an AI’s responses depends not just on which model you use, but on what it remembers and what information it can retrieve at the right moment,” said Junha Jang, CEO of Celon. “Memory aims to connect conversations, policies, code, and search results that are scattered across multiple chats and services, so that AI can find better, understand better, and ultimately give better answers.”