Sogang University Professor's AI Startup 'INT.' Selected for Government's TIPS Program

Professor Park Hyun-kyu from Sogang University's Graduate School of Technology and Management has successfully launched his startup, INT., which has been selected for the government's '2026 TIPS General Track Program.' This selection grants INT. approximately 1.1 billion won (about $800,000 USD) in research and development funding over the next two years.

Significantly, INT. achieved this milestone just 79 days after its official incorporation. The TIPS (Tech Incubator Program for Startup) program is a key public-private initiative where the government supports R&D for promising tech startups identified and invested in by private venture capital firms. INT. had already secured investment commitments and a TIPS recommendation from CNTTech Accelerator, a major TIPS operator, prior to its official establishment, based on its recognized technological capabilities and business potential.

INT. is developing 'Turing,' an AI evaluation automation solution. This tool is crucial in an era where AI is evolving into 'digital labor' that autonomously performs tasks. Turing addresses two primary issues in enterprise AI: the difficulty in accurately verifying domain-specific performance during the development phase and the challenge of continuously monitoring and rapidly responding to performance degradation during the operational phase.

“

For the past four years, our lab's goal has been to 'develop and supply AI that is actually used in the field.' We have successfully completed 19 corporate projects.

Unlike general benchmarks like MMLU, Turing provides developers with evaluation metrics tailored to the client's actual business situations. This prevents the creation of AI that scores high on benchmarks but fails to perform tasks effectively in real-world applications. During development, Turing uses domain-specific metrics such as Keyword Error Rate, Hallucination Rate, and Faithfulness to verify AI performance. In the operational phase, Turing's multi-agent system automatically sets evaluation criteria based on the client's domain, continuously measuring metrics like Intent Recognition Accuracy, Task Success Rate, and Negative Flip Rate. It also diagnoses performance degradation, proposes recovery plans, and implements them with minimal service impact, eliminating the need for constant manual system maintenance.

Professor Park, CEO of INT., stated, "For the past four years, our lab's goal has been to 'develop and supply AI that is actually used in the field.' We have successfully completed 19 corporate projects." He attributed this "small but meaningful achievement" to the researchers and staff who "steadfastly believed and worked together through the challenging process." INT. is pursuing integrated enterprise AI construction projects and the operation and management of these systems using Turing. They have already built enterprise AI for major corporations like LG CNS, various small and medium-sized enterprises in sectors such as content and healthcare, and public institutions like the Science and Technology Policy Institute (STEPI). Park believes the competitive focus in AI is shifting from 'creating AI' to 'effectively operating adopted AI,' aiming to preempt the AI operation market, which he sees as not yet fully competitive.

“

This small but meaningful achievement was possible because of the researchers and staff who steadfastly believed and worked together through the challenging process.