Works above the existing EMR
The initial workflow does not require an EMR API modification or a system replacement.
Workflow-specific configuration and validation are still required at each site.HayanMind presents FreeDoc
Local-first AI for medical claim administration.
FreeDoc works above the EMR clinics already use. It reads the active screen, checks relevant claim rules, and supports the next action while keeping clinicians and administrators in control.
- Existing EMR
- Local-first
- Human approved
Ready for review
Screen understood
Rule check queued
Synthetic-data demonstration No real patient record is shown.
The administrative gap
Claim administration still depends on fragmented screens and manual rule checks.
The work is operational, repetitive, and highly specific to each site. Staff move between legacy interfaces and changing claim guidance—then review the same details again before submission.
- 01Different interfaces
Legacy EMR screens and workflows vary between providers and sites.
- 02Changing criteria
Insurance claim rules are detailed and updated frequently.
- 03Repeated handling
Lookup, entry, review, and correction consume administrative time.
working day per week
Claim review was described as a recurring burden of roughly one working day per week in an interview with a partner clinic. This is an interview finding, not a measured cross-clinic average.
How FreeDoc works
Same screen.
Different operator.
FreeDoc follows the same visible workflow as a human operator, with the person responsible for the final approval.
- 01Capture
Read the active EMR screen.
- 02Understand
Identify the relevant fields and actions.
- 03Check
Retrieve applicable rules and test conditions.
- 04Act
Use mouse and keyboard inputs on the existing interface.
- 05Approve
Keep a human in the approval loop.
Boundary FreeDoc supports claim administration. It is not presented as an autonomous system for clinical judgment, diagnosis, or treatment decisions.
Why FreeDoc
Designed around the system already in place.
Local-first architecture
Core model inference and screen control are designed to run on a dedicated device inside the clinic.
Access, logging, retention, masking, and security controls must be configured and validated with each partner.Human-approved workflow
FreeDoc supports the next administrative action while keeping clinicians and administrators responsible for approval.
It does not automate clinical judgment or represent a completed regulatory certification.Technology
Reliability is built around the model.
SmartGrounding combines visible text, multimodal prediction, verification, calibration, and recovery into one screen-action pipeline.
- 01OCR MatchAnchor visible labels
- 02VLM PredictEstimate the target
- 03Self-checkVerify the intent
- 04CalibrateCorrect coordinate drift
- 05RecoverDetect and retry failure
Internal sandbox
From a single-model baseline to a system-level result.
Each additional check is designed to catch a different class of screen-grounding failure.
23.2%Single-model baseline
→
92.9%90 of 97 tasks
Internal SmartGrounding sandbox result: 90 of 97 tasks. This is not clinical performance, claim accuracy, certification, or a production deployment result.
47,798Disease-code entries in the current Korean knowledge base
76Drug–disease mapping rules
150+Deduction-risk patterns
7B / ~8GBPrototype model class / inference memory
These counts describe the current Korean project knowledge base and prototype. They do not imply global claim-rule coverage.
Current readiness
A working workflow, in a controlled scope.
The current prototype demonstrates screen-based tasks in a repeatable EMR simulation using synthetic data. The next step is not broad deployment—it is a limited, workflow-specific validation.
- Demonstrated
- Patient search, chart opening, screen-element selection, and claim-form interactions.
- Environment
- Controlled EMR simulation with synthetic data.
- Feasibility
- Paid feasibility PoC with a dermatology MSO partner, March–May 2026.
- Next
- Limited validation covering site-specific exceptions, approval steps, access permissions, logs, and security controls.
Controlled prototype
Partnership
Let us validate one administrative workflow together.
We are looking for partners who can define the real operating context and evaluate FreeDoc against shared workflow, security, and approval criteria.
- Clinic, hospital group, or MSO
- EMR, HIMS, or RCM vendor
- Healthcare IT distributor or implementation partner
- 01
Select one workflowStart with a single, real administrative task.
- 02
Define boundariesMap exceptions and human approval points.
- 03
Confirm local conditionsReview regulatory, privacy, and security requirements.
- 04
Run in shadow modeValidate within a limited scope before operational submission.
- 05
Measure togetherTrack agreement, handling time, exceptions, feedback, and security findings.
Singapore is a validation step—not a copy-and-paste rollout. Local workflow fit and operating requirements need to be established with local partners first.
A KAIST-rooted software and AI team, founded in Daejeon in 2017.
HayanMind brings software product development, global operations, and applied AI experience to FreeDoc. The operating record below belongs to the company—not to the product’s deployment history.
180countries reached by HayanMind consumer services
3Mcumulative users and downloads
300K+monthly active users at scale
>80%of 2024 revenue generated overseas
KRW 847M2024 revenue · profitable in 2024
KRW 3.267Bcumulative equity investment
These figures describe HayanMind’s software operations, not FreeDoc deployment.
A narrow first pilot
Let us validate one administrative workflow together.
A narrow first pilot can establish workflow fit, local data control, measurable operating value, and the right boundaries for broader deployment.
Discuss a Pilot
Jay Oh
CEO, HayanMind Inc.
jmoh@hayanmind.com
Do not include patient or medical information in an email to us.