AI does not replace physicians.
It brings evidence to every decision.

MEV is an AI medical agent company serving tertiary and university hospitals. We support three functions — reimbursement and claims optimization (RCM), automated clinical coding, and referral and return-referral management — inside the clinician's existing EMR/OCS workflow. Every recommendation is accompanied by evidence, and every decision is signed off by a physician.

MEV's agent is not a single-purpose diagnostic aid. It supports the entire administrative workflow from documentation through coding, claims, and referrals. It generates recommendations and drafts inside the EMR/OCS screens, with no separate system to adopt, and attaches a source, physician sign-off, and an audit trail to every output.

Before submission, MEV screens each claim against the review criteria for items at risk of denial and for missed charges, and shows the evidence for why each item is flagged. It also monitors HIRA quality-assessment metrics — including the New Bundled Payment system, DRG, and non-covered services — to protect rightfully earned revenue.

From clinical documentation, imaging reports, and procedure notes, MEV recommends diagnosis (KCD), fee-schedule, and procedure codes, and displays the supporting text from the record alongside each recommendation. The AI only recommends. The physician makes the final decision, and finalized codes flow directly into EDI claims.

It automatically identifies patients ready for return referral, drafts referral and return-referral letters, and matches patients with partner facilities. It supports management of the return-referral rate, a key metric in Tertiary General Hospital designation review.

No. MEV follows a human-in-the-loop principle. The AI presents recommendations and drafts together with their evidence, and the final decision always rests with the physician. Who approved what, and when, is recorded in the audit log.

Patient data never leaves the hospital network. MEV supports on-premises and air-gapped (network-separated) deployment, pseudonymizes data before processing, and transmits no identifiable information externally.

MEV is designed to the requirements of the Medical Service Act and the Personal Information Protection Act from the outset, with built-in procedures for verifying the clinical validity and evidence of every output. We also assess applicability under the MFDS medical device (SaMD) framework.

MEV is not a stand-alone application. It operates inside the EMR/OCS screens your clinicians already use, presenting recommendations without a separate login or duplicate data entry, and without interrupting the flow of care.

In a 30-minute demonstration, we walk you through the area that matters most to you — claims, coding, or referrals. Measured outcomes from client hospitals will be published separately once the data has been validated. We do not present estimates as facts.

An AI medical agent is an AI system designed to carry a specific hospital workflow (patient referral and return referral, insurance claims, clinical coding) through to completion inside the clinician’s existing EMR/OCS workflow. Unlike a chatbot or a single prediction model, it reads the record, produces drafts and recommendations, attaches evidence, hands off to the next system after physician sign-off, and logs the entire process in an audit trail. MEV’s agents stop at recommendations and drafts. The physician always makes the final decision.

It is the standards layer between hospital systems (EMR/OCS, billing, medical records, referral exchange systems) and the AI agents. It handles HL7 FHIR-based EMR/EHR integration, KCD ↔ ICD-10 and SNOMED CT terminology mapping, country-specific claims adapters, and on-premises LLM operation with security controls (pseudonymization, audit trail, physician sign-off). Once integration, terminology mapping, and security review are completed at first deployment, each subsequent agent simply plugs into the same foundation.

The middleware is designed around HL7 FHIR interfaces and KCD ↔ ICD-10 and SNOMED CT mapping. Care transition, claims and denial management (RCM), medical coding (ICD-10 and CPT), prior authorization, and clinical documentation are on the global expansion roadmap. Target markets and timing will be confirmed according to the business plan. Commercial deployments today are in Korean tertiary hospitals.

Seoul National University Bundang Hospital began a subscription to the referral and return-referral AI agent in October 2026 (SaaS subscription, contracted through the hospital’s project lead vendor). Proofs of concept are in progress with The Catholic Medical Center (reimbursement and claim-denial response AI) and Seoul St. Mary’s Hospital (health-screening report generation AI). Measured outcomes will be published once the data has been validated.