[For more diverse company information on startups, visit Unicorn Factory's big data platform Data Lab.]
"Not every company can have AI (artificial intelligence) experts in-house. For companies whose core business is not technology, it is far more advantageous for growth to focus on the essence of their best-performing businesses by using data and AI as tools rather than investing resources in building AI systems."
Park Kyung-ho, CEO of Clozer Labs, said, "Just like 'closer' refers to a relief pitcher who finishes games in baseball, we founded this company with the goal of creating AI that neatly solves business problems faced by clients and boosts productivity."
The name Clozer Labs carries not only the meaning of a closer but also the intention to bridge the gap between the thoughts of corporate decision-makers and actual on-site data, helping them make decisions that are 'closer' to better conclusions.
Clozer Labs is a startup established in March 2024 by experts from the AI team at SOCAR, a mobility platform company, including Park Gyeong-ho (CEO). Park (CEO) worked on AI tasks at SOCAR aimed at reducing vehicle management costs and conducted research and development on customer consultation projects utilizing AI.
Park (CEO) stated, "As a result of our joint research with team members at the time, we concluded that developing services to replace human consultants is difficult given the current level of AI. We became convinced that it is better to focus on work assistance that helps people perform their tasks faster and more efficiently rather than replacing them."
Clozer Labs initially started as a data-based customer consultation (CS) solution that recommends appropriate responses to consultants and provides real-time customer information. It has since expanded into an AI-driven decision-making platform applicable across the entire organization, including marketing, sales, and product planning.
Standardizing fragmented corporate data… also resolving hallucinations
The flagship product, 'Deskroom,' automatically standardizes unstructured data scattered within companies and aids decision-making. The company explains that AI automatically refines and formats various forms of corporate unstructured data—such as orders, reviews, social media (SNS) responses, customer inquiries, and internal reports—into analyzable standards.
The core lies in its proprietary 'Ontology' technology. Ontology refers to a knowledge structure that systematically defines concepts within a specific subject and the relationships between them. In other words, it is like pre-drawing a map so that AI can understand not just what data means but how things are connected, rather than merely storing data.
Park (CEO) said, "Deskroom integrates fragmented data to create a foundation where large language models (LLMs) or AI can interpret data according to the same standards as humans, enabling AI to generate quantitative metrics tailored to each company's unique criteria."
It also resolves hallucinations, one of the biggest issues with AI-based solutions. He explained, "In areas involving numerical answers or analysis where there is a correct answer, we limit the AI's scope (by setting schemas) to provide accurate values without errors. In areas requiring value judgments, we recommend optimal options to humans."
Schema setting means, for example, when analyzing a photo of vehicle damage, limiting the response area to focus only on 'whether the bumper is damaged' rather than the vehicle's color or model name. Park (CEO) emphasized, "In numerical areas with correct answers, we boast an error rate of less than 1%, effectively close to 0% accuracy."
"Useful for companies where quality control is critical, such as manufacturing, beauty, and defense industries"
Deskroom also boasts rapid deployment speed. Park (CEO) stated, "We already have established work standardization criteria and know-how for each industry sector. While traditional system integration (SI) companies may take several months to complete tasks, we finish them in as little as two weeks or at most two months."
Unlike dashboards that simply display past data, Deskroom's differentiated competitive edge lies in providing forward-looking analysis and insights such as 'what should be done now' to improve corporate operations. Additionally, it is designed for easy use even by non-AI experts.
Major clients that have adopted Deskroom include beauty company APR, food-tech company Ignis, and education platform Class101. Park (CEO) said, "We are supplying solutions to companies operating under 'multi-product, small-batch production' systems where quality control is critical. Recently, we have also expanded into the defense domain."
He added, "Adopting companies provide feedback that they can now focus solely on their core business—product planning and marketing—without needing data or AI experts. Overall, clients evaluate that Deskroom has improved efficiency by approximately 10% of their working hours."
Park (CEO) noted, "While the term 'AI agent' is commonly used in the market, we are targeting something more specific: an 'AI-driven decision-making platform (ADP, AI Driven Decision Making Platform).' Companies like Palantir in the U.S. have demonstrated the value of this area, and the market is growing rapidly."
He further stated, "This year, we plan to enter new vertical markets with abundant unstructured data, such as facility management and construction. I believe that The Korea Economic Daily's core driver for further growth lies in data and AI. We aim to become a partner that helps non-tech companies avoid being left behind due to data and AI."
[MoneyToday startup media platform 'Unicorn Factory']
in FUTURE
