Omelet: “Prioritizing Logistics Optimization with ‘Oasis’ AI to Support Corporate Decision-Making”

Omelet. As a French-style omelet, it’s a somewhat unexpected name for a startup. When asked why they named the company that way, Omelet responded, "We combined the middle letters of Optimization and MachinE LEarning Technologies." It’s a word that combines 'optimization and machine learning technology,' and it was an idea from the company members. It’s a slightly cute and memorable name.

However, Omelet is not soft. On the contrary, it stands at the forefront of optimization AI technology. The collection of optimization AI technologies they have created is called 'OaaSIS.' It stands for 'Optimization as a Service/Intra/System,' which clearly shows what Omelet aims for. It’s not just a service but extends to the optimization of infrastructure and systems. What does this mean?

Omelet first applied OaaSIS to logistics systems. When you access OaaSIS's logistics optimization system, it clearly shows which route to take to achieve the best results. In this process, you can prioritize 'time reduction,' 'even distribution among delivery drivers,' or 'distance,' among other conditions. Of course, multiple conditions can be applied, and even weights can be assigned to each condition.

But can traditional AI methods, which derive results by learning from past data, achieve optimization? Such learning methods can only follow the optimal results created by past workers. Therefore, OaaSIS was created as a system that goes beyond learning from past data and generates data on its own to learn. It’s similar to how AlphaGo initially learned from human Go players' records but later generated tens of millions of records on its own to learn.

Omelet CEO Park Jin-kyu often uses the term 'solution' from mathematics instead of the commonly used term 'inference' in AI technology. It means finding a new 'solution' in a state where no answer exists. "OaaSIS is about finding new 'solutions' through the process of continuously improving and optimizing the best answers created by humans, that is, through reinforcement learning," emphasized CEO Park.

Based on the OaaSIS created this way, Omelet plans to release a routing engine for delivery allocation and route optimization in the form of an API worldwide. In other words, they aim to make customers around the world access the Omelet system and use the routing engine. It’s an ambitious plan.

This alone is impressive, but what does it mean to expand OaaSIS beyond just a service to infrastructure and systems? CEO Park Jin-kyu revealed his ambition, saying, "OaaSIS will start with logistics systems but will expand its application range to traffic systems, unmanned robot operation automation, semiconductor chip design, and new drug candidate development." It means they aim to develop it into a kind of artificial general intelligence (AGI). "We have the core technology of optimization AI. We are a team that can develop foundation models and provide them to other teams," CEO Park expressed his pride.

Looking at the composition of the Omelet team, such ambition and pride are well reflected. CEO Park Jin-kyu and CTO Kwon Chang-hyun are both professors at KAIST's Department of Industrial and Systems Engineering, and the team members are also composed of master's and doctoral graduates from KAIST and Seoul National University. Omelet has continued to achieve results, securing seed investment three months after its establishment and being selected for the deep-tech TIPS program shortly thereafter.

Omelet has been selected for the 7th cohort of 'O!VentUs (Open + Venture + Us),' an open innovation program jointly operated by CJ Group and the Seoul Center for Creative Economy & Innovation, receiving support and conducting PoC for logistics systems with CJ Logistics and various companies.

It’s not just the reporter who is curious about whether the day will soon come when people around the world access OaaSIS and use the engine of the Korean company Omelet, just as we access OpenAI's ChatGPT or Google's Gemini server and use their engines.

What problems is Omelet trying to solve?

Omelet aims to solve the difficulties companies face in making effective and economical decisions using vast and complex data. As the complexity of problems increases, traditional methods have reached their limits, and the labor costs of optimization experts and IT personnel, as well as the costs of building and maintaining models, have skyrocketed. Moreover, generative AI has limitations in deriving effective decisions, and in industrial settings, there is a lack of data necessary for decision-making, and astronomical costs are incurred in building and operating AI. Omelet aims to solve these problems through optimization AI technology, supporting companies to make more efficient and economical decisions.

How do you solve this problem?

Omelet provides the optimization AI service 'OaaSIS (Optimization as a Service/Infra/System)' to help companies economically derive effective decisions.

We supply optimization AI algorithms, not just passive AI that performs predictions, but active AI models that receive problem data, generate solutions to the problem, and continuously improve. This optimization AI model can derive decisions based on the data companies have and also has the ability to generate data on its own to learn, solving larger problems faster and more accurately.

Omelet also provides domain-specific decision-management software to support customized decision-making solutions tailored to the needs of each industry. This allows companies to derive optimal solutions for complex problems and build AI models economically without relying on company data.

Furthermore, Omelet offers the DecisionOps platform, which allows companies to quickly build, maintain, and manage AI-based decision-making models. This platform continuously learns based on data, automatically updates AI models as new data accumulates, and redeploys decision-making models to maximize decision-making efficiency. Companies can automate complex decision-making more simply and economically.

What are our competitive advantages and technical strengths compared to competitors?

Omelet's technical strength lies in possessing core optimization AI technology that solves complex decision-making problems using artificial intelligence. Omelet's optimization AI model is an active AI that generates and continuously improves optimal solutions in response to problem data, unlike existing passive AI models. This allows for more effective problem-solving as it can explore better solutions.

Additionally, Omelet's technology does not rely on company data but learns using self-generated data, solving larger problems faster and more accurately. This significantly reduces the costs associated with data collection and model training for companies, providing an economically efficient solution.

Omelet's optimization AI, similar to OpenAI's recently released O1 model, has the ability to perform repetitive inference and calculations based on previous results during testing to derive better outcomes. While the O1 model presents the possibility of solving complex problems in various fields such as mathematics, engineering, and science, high development and operational costs remain a significant challenge. In contrast, Omelet has the technology to enhance performance and reduce development costs by fine-tuning the Test time inference module for each problem, providing a more practical and economical solution.

What services does Omelet provide?

Omelet develops various solutions to solve complex problems in logistics, robotics, and transportation using optimization AI technology. Currently, Omelet is preparing to launch major product lines aimed at optimizing logistics system operations.

Firstly, we plan to release a routing engine in the form of an API worldwide that can solve delivery allocation and route optimization problems. This routing engine combines generative AI and AI-based Local Search techniques, reflecting industry-specific constraints and solving large-scale routing problems quickly and accurately. Additionally, a modular structure is applied, allowing flexible adaptation to various industries with minimal additional work.

Secondly, the Transportation Management System (TMS), scheduled for release in November, supports delivery optimization through hierarchical routing that considers resource allocation between logistics centers. It efficiently manages customers' delivery data through intuitive optimization settings and integrated data management functions, and easily handles service times and entry conditions. Furthermore, TMS can be used as a CRM tool to manage delivery quality and customer satisfaction.

In addition, we plan to release logistics robot task allocation and route optimization engines and software utilizing them next year.

Who are our target market and key customers?

Omelet's target market includes various industries that require AI-based optimization solutions, particularly focusing on logistics, transportation, robotics, and manufacturing, which require large-scale resource management and complex decision-making. The market size is in the billions of dollars globally for logistics and transportation management solutions alone, and these industries are seeking to reduce costs and improve efficiency by adopting AI and optimization technologies.

Key customers are large and mid-sized companies that need to solve complex resource management and operational optimization problems, particularly logistics companies, transportation and delivery service providers, and manufacturers. These companies can make better decisions quickly and economically using Omelet's optimization AI solutions, maximizing operational efficiency.

What is our business model?

Omelet's business model consists of three components: SaaS for optimization AI algorithm supply, SaaS for software supply, and PaaS for the DecisionOps platform.

First, the algorithm supply SaaS. Omelet provides optimization algorithms for solving complex problems in the form of APIs. API usage fees vary depending on the scale, complexity, and frequency of calls. We offer customized solvers and consulting services for large-scale enterprises and share a portion of the cost savings achieved through algorithm usage as revenue through a performance-based fee structure.

Second, the software supply SaaS. Monthly or annual subscription fees are charged based on the number of users, with additional fees applied for premium features beyond the basic functions. We offer customized software packages and consulting services for large-scale customers.

Third, the DecisionOps platform PaaS. Monthly or annual subscription fees and API usage fees support customers in building decision-making pipelines. Additionally, we operate a marketplace within the platform where decision-making models can be traded, charging a commission on traded models.

What achievements has the Omelet team made so far?

Omelet secured seed investment three months after its establishment in 2023 and was immediately selected for the deep-tech TIPS program, securing 1.7 billion KRW in research and development support over three years. We have conducted PoC (Proof of Concept) projects with over 10 mid-sized and large companies, confirming the industrial applicability of optimization AI.

In the second half of 2024, we plan to complete the development and launch of the logistics routing engine API, driver app, and TMS software, officially supplying the logistics delivery optimization system to the market. We have achieved the compressed development of core optimization AI technology and various applications based on it in a short period.

What is our team's competitive edge?

Omelet's greatest competitive edge lies in its team composed of highly skilled individuals based on world-class optimization AI technology. Founded by CEO Park Jin-kyu and CTO Kwon Chang-hyun, both current professors at KAIST, Omelet consists of master's and doctoral graduates from KAIST and Seoul National University with world-class research achievements in the optimization AI field. Especially, Omelet has recruited developers with rich industrial experience and domain experts, combining technology and practical experience, and currently, five KAIST students are actively contributing as interns.

Our team aims not just to stay at the level of technology but to provide innovative solutions that can bring about substantial changes in the industry. All team members are united under the same goal, solving complex and challenging problems, and have a strong desire to leap into a global big-tech company. Through this high density of talent and grand vision, Omelet is creating innovation faster and more efficiently than any other company.

Three reasons why we should receive investment!

First, Omelet has a team with a high density of talent composed of master's and doctoral graduates from KAIST and Seoul National University. Led by CEO Park Jin-kyu and CTO Kwon Chang-hyun, who have world-class research achievements, the team includes developers with rich industrial experience and domain experts, all with a strong desire to leap into a global big-tech company.

Second, Omelet possesses core technology in the optimization AI field and boasts the technical capability to actively solve complex problems. We are developing a routing engine and TMS software applicable to various industries such as logistics, transportation, and manufacturing, providing innovative solutions that are economically scalable.

Third, Omelet has a strong vision for AGI technology to overcome the limitations of generative AI in the current industrial field. This presents new possibilities for companies to make more efficient decisions and bring about innovation across industries.

What help did you receive through the CJ O!VentUs program?

It was very meaningful to have the opportunity to solve important real-world problems with CJ Logistics. It was a valuable opportunity for a startup to prove that it could solve significant problems in actual industrial settings, beyond just the problems it set for itself with its own technology. Through this collaboration on 'delivery box size optimization recommendations,' we not only secured an opportunity to supply Omelet's solution but also instilled confidence and pride in our team, reinforcing their belief that Omelet’s technology can make a significant industrial impact.

Any additional messages you would like to convey?

The OaaSIS optimization AI technology being developed by Omelet is a model with AGI attributes that can actually solve complex problems, unlike many AI models that merely perform predictions. As there are still few companies challenging this field, we believe we are one of the few companies in Korea developing core AGI technology.