LG CNS and Cohere advance enterprise agentic AI with on-premise security focus

LG CNS and Canadian AI firm Cohere launch AgenticWorks, a secure, no-code platform designed to bring autonomous, customised AI solutions to Korea’s corporate sector, prioritising data sovereignty and operational control amidst industry challenges.

LG CNS is making significant strides in the development and deployment of agentic artificial intelligence (AI), a next-generation form of AI capable of autonomous judgement and action without solely relying on input-output operations typical of existing generative AI. This technology is poised to revolutionize productivity in the business-to-business (B2B) sector, marked by expectations of transformative operational efficiencies. However, companies aiming to adopt such agentic AI face substantial challenges, particularly regarding security, trust, and control. This is especially critical in sensitive industries like finance, manufacturing, and public sectors where data privacy is paramount and where many firms prefer on-premise AI deployments over cloud-based solutions.

To address these concerns, LG CNS has partnered with Cohere, a Canadian AI firm renowned for its focus on enterprise AI and efficient model architectures. Together, they have launched the AgenticWorks platform, designed specifically for corporate use with an emphasis on security and customization. AgenticWorks offers a no-code development environment, enabling companies to build AI services without programming expertise, while leveraging Cohere’s Korean language-specialized large language model (LLM). This platform emphasizes fine-tuning models on company-specific data within secure, on-premise infrastructures, enhancing the AI’s contextual understanding and reliability. Lim Eun-young, head of LG CNS’s AI Centre, highlighted in an interview the platform’s key advantage of enabling rapid, tailored agent development that meets stringent customer requirements, particularly in secure environments.

The collaborative work with Cohere represents a strategic choice for LG CNS to differentiate itself from major global AI providers such as OpenAI and Google, which often rely heavily on public cloud ecosystems. According to Lim, although the AgenticWorks platform can operate in public cloud environments, LG CNS’s competitive edge lies in delivering equally advanced technology securely on-premise—an attractive proposition for customers prioritizing data sovereignty and control. Cohere further asserts the efficiency of its AI architecture, claiming that their models can achieve comparable benchmarking performance with markedly fewer GPU resources compared to competitors, thereby offering cost-effective scalability.

South Korea’s market holds particular appeal for Cohere owing to its highly educated, digitally native population and proactive government initiatives around AI sovereignty, remarks Ivan Zhang, Cohere’s co-founder. Security and privacy are foundational to their design philosophy; their systems are engineered to prevent data leakage and function seamlessly offline, particularly in on-premise deployments. The platform’s training approach—employing retrieval-augmented generation (RAG)—helps minimize AI hallucination, a common issue where AI generates false information, by ensuring responses are grounded in transparent source references.

Recognising that fully autonomous AI managing mission-critical tasks is not yet feasible, both LG CNS and Cohere stress the continued necessity of human oversight in AI decision processes. This cautious approach is seen as essential for maintaining operational integrity.

Building on its Cohere partnership, LG CNS has also allied with Silicon Valley-based AI firm Weights & Biases (W&B) to further enhance agentic AI capabilities. This collaboration aims to improve the performance and scaling of AI services offered to enterprise clients, integrating W&B’s tools—such as ‘Weave’ for monitoring AI model performance and ‘Models’ for training management—to optimize operations. These strategic partnerships underscore LG CNS’s ambition to secure a leading position in Korea’s emergent agentic AI market.

Currently, LG CNS’s agentic AI efforts are focused on proof of concept projects mainly within the financial sector, addressing clients’ needs for secure and reliable AI solutions. However, the company envisions expanding into healthcare, public administration, manufacturing, and other industries where secure, trustworthy AI is indispensable. Lim pointedly observes that for organisations to truly harness the benefits of agentic AI, they must be prepared to innovate their entire operational processes, even undertaking systemic changes if required.

As the agentic AI market in South Korea is still nascent, LG CNS’s integrated approach—emphasising security, localisation, customisation, and operational optimisation—positions it at the forefront of this fast-evolving technological frontier. The combined expertise of LG CNS, Cohere, and W&B aims to catalyse a new era of AI-enabled enterprise innovation, carefully tailored to meet local regulatory and business expectations.

📌 Reference Map:

• – Paragraph 1 – ^[1], ^[4]
• – Paragraph 2 – ^[1], ^[2], ^[4]
• – Paragraph 3 – ^[1], ^[4]
• – Paragraph 4 – ^[1], ^[4]
• – Paragraph 5 – ^[1], ^[4]
• – Paragraph 6 – ^[1], ^[3], ^[5], ^[6], ^[7]
• – Paragraph 7 – ^[1]

Source: Noah Wire Services