[DigitalToday reporter Seulgi Son] In South Korea, more companies are trying to introduce Forward Deployed Engineer (FDE) organisations to apply generative artificial intelligence to real work.

FDE is a field-focused engineering model established by U.S. data platform company Palantir in the course of carrying out projects for defence and intelligence agencies. Engineers enter a customer organisation directly, define problems and build tailored AI systems. In South Korea, it is used not only by deploying teams to client sites but also as an organisational form that supports internal AI transformation (AX).

Naver Cloud launched a defence AX task force (TF) as of the 1st and highlighted its FDE organisation. The TF is headed directly by Naver Cloud CEO Kim Yoo-won (김유원). The company’s ambition is to find a business opportunity with sovereign AI in the defence sector, where security concerns limit the use of foreign cloud services and AI models.

FDEs are deployed directly to defence sites to implement AX in military organisations. Naver Cloud in particular expects its own omnimodal AI model to be highly usable in battlefield environments. It judges that a structure that integrates text, images and audio in a single model is suited to battlefield settings where voice communications, video information and documents move together. It also cites having an in-house full-stack capability in South Korea, from cloud to AI models and services, as a competitive edge over foreign solutions.

A Naver Cloud official said, "In the defence sector, there are limits to using foreign AI or cloud services due to data sovereignty and security issues, so we expect the competitiveness of providers that can offer both domestic AI models and infrastructure to stand out," adding, "We are currently at the stage of identifying related business opportunities."

Physical AI startup Makinarocks has operated an FDE organisation since 2024. FDEs are deployed to customer sites and collaborate with domain experts and the client’s IT organisation, taking responsibility for the entire process from AI system design to operations.

Makinarocks previously built industrial-site-specific AI solutions for major South Korean manufacturing conglomerates such as Doosan, Samsung, Hyundai, LG and SK. Dispatched engineers develop and deploy industrial AI models such as anomaly detection, optimisation and predictive analytics, and the company’s machine learning operations (MLOps) platform, Runway, is used to integrate management from model development through operations.

Makinarocks CEO Yoon Sung-ho (윤성호) said, "From battlefields to factories, we aim for technology and an organisation that takes responsibility for the entire process until AI operating in the real field translates into results."

More companies are also trying to use FDEs for internal AX.

Kakao created an FDE organisation around April to support internal AX. Some senior engineers in its existing AI-native organisation moved into the FDE team. FDEs rotate across roles regardless of whether work is development or non-development, analyse work processes and support the introduction of AI agents and automation.

A Kakao official explained, "To embed AI into real work, there needs to be a process of understanding on-the-ground problems and directly making and applying the tools needed," adding, "The FDE organisation is playing a role in connecting that."

Krafton has also recently started running FDEs for internal AX. It conducted hiring for an AI FDE role in March and selected a double-digit number of people. FDEs are assigned to teams that need AX and use coding agents to design, develop and deploy required programs. To secure strong engineers, Krafton offered compensation including base pay above the starting salary for regular employees and performance bonuses, with hiring regardless of career history or academic background.

A Krafton official said, "It is a structure where FDEs solve the AX tasks we throw at them, and members’ reactions are very good," adding, "We believe the importance of roles like FDEs that understand on-the-ground problems and can implement them with AI technology will grow further."

Some point out that the FDE model being tried in South Korea is not fundamentally very different from existing system integration (SI) businesses. The analysis is that for AI businesses led by FDEs to secure sustainability, competitiveness in the AI solutions and platforms they build must support the effort, rather than relying on dispatched personnel themselves.

An AI solutions industry official said, "Korea is originally a market with highly capable SI engineers, so there are strengths even in the FDE approach of going into client sites and solving problems," adding, "Under the current system that constrains performance-based compensation for engineers, it is difficult to continuously secure highly skilled engineers and grow the ecosystem."

Another AI infrastructure industry official said, "In the United States, it is a SaaS-centred market so FDEs were accepted as a relatively new role, but in Korea, a culture of dispatched engineers is already established," adding, "For FDEs not to be consumed as a simple marketing term, differentiated AI models and platform competitiveness must be proven together."