Kakao selected as operator for 11 billion won AI Agent Marketplace development project by South Korean government
The AMW Read
Kakao secures a government-backed AI agent marketplace project, an incremental development for a known player in the agent space, but with segment-level implications for Korea's agent ecosystem.
Kakao selected as operator for 11 billion won AI Agent Marketplace development project by South Korean government
South Korea's Ministry of Science and ICT, through the National Information Society Agency (NIA), has selected Kakao as the lead operator for an 11 billion won (~$8.2 million) AI agent marketplace development project. Kakao will partner with its enterprise subsidiary Kakao Enterprise in a consortium, with work running through December 2027. The project combines government funding with private contributions, and the consortium aims to deliver a first public release within the year, followed by a second release in 2027 that includes an agent studio.
The marketplace is designed as an open architecture that is not locked to any specific cloud or AI model. Beyond Kakao's own Kanana model, the platform will integrate domestic and international models, with external APIs and Model Context Protocol (MCP) enabling interoperability between tools. Registered AI tools will undergo security review, sandbox validation, and hallucination controls before publication, with ongoing re-verification during operation. Kakao already operates PlayMCP, an open platform hosting around 400 MCP servers, and provides an AI national assistant pilot with the Ministry of the Interior and Safety.
The project signals South Korea's push to create a domestic hub for AI agent distribution and validation, rather than relying on global platforms. For builders, the government backing provides a funded channel to reach Korean enterprise and public-sector users, with the consortium planning an AI agent challenge and hackathon to recruit and onboard promising agents. The emphasis on security review and sandbox validation suggests that compliance-ready agents will have a smoother path to distribution in this marketplace, potentially setting a template for how other governments structure AI agent ecosystems.
