Metanet said on Tuesday it held an in-house AI idea contest, the AI Agent Hackathon, for employees.

The company set up the contest as part of its strategy to strengthen applied AI competitiveness by advancing how employees use AI and accelerating the embedding of AI capabilities across the organisation.

Metanet Global used its no-code and low-code AI agent work automation platform, Meta AI Studio (Meta AI Studio), to provide an environment where non-developers can build AI agents through drag-and-drop. Employees received support including training and solutions to design work processes themselves and build AI agents linked to email, documents and work systems.

The company said 15 teams with a total of 60 people attended the hackathon, and six teams reached the finals after five weeks of pilot development. They built and demonstrated prototypes on topics including handling repetitive tasks based on AI agents, automated email briefings, automating work evaluations and automating quotation calculations. Evaluations focused on technical merit, business scalability and the level of work automation.

As a result, the requirements-based automated quotation AI agent solution SMART (Software Measurement & AI-driven Requirements Tribe) received the highest scores for technical merit and business scalability and was selected as the overall winner.

SMART automates the entire process with AI, from estimating project scale to calculating quotations and drafting quotation documents. It cut quotation processing time by about 80 percent, from 40 hours a week to 8 hours, enabling work that previously took a week to be completed within a day, the company said.

Kwang-soo Lee (이광수), a director in Metanet's xAIOps Group who led the SMART project, said it applied estimation methods based on international and domestic standards including IEEE 29148, IFPUG and COCOMO2 (KOSA) to improve objectivity and reliability in quotation calculations. He said a strength was that it was built to allow results to be documented, stored and reused. Metanet is reviewing ways to apply top projects derived from the hackathon to actual work.