Coretrustlink

FOTYPro

Internal Partner that Protects

Corporate Trust

Automatically managing complex

regulations and personal

information risks, reducing costs

and increasing trust

Key Features

LLM-based Document Analysis

Applying corporate-tailored personal information protection through

context-based analysis,

not keyword-based

Handling both structured and

unstructed data and applying

to varius industries

(e.g., law, finance, insurance)

Personal Information

&Leakage Prevention

Detection of sensitive and

confidential data in internal

corporate documents and DB

Prevention of unauthorized

attemps to leak transaction data, customer consent, etc

Efficient Utilization

On-premise/cloud-based

distribution is available

Providing an intuitive web-based

UI/UX for search results,

log tracking, etc

Providing masking exception

process and function to

print after saving masked documents

Key Technologies

MAS

(Multi-Agent System)

Distributed Autonomy
Each agent makes independent decisions, allowing the system to continue operating partially even in case of failures.

Collaborative Problem-Solving
Multiple agents can exchange information and divide roles, enabling distributed handling of complex internal control and risk management tasks.

Scalability
New functions or modules can be easily integrated by attaching them to agent units, making expansion flexible.

RAG AI

(Retrieval-Augmented Generation)

Efficient Maintainability
Adding new data to the index alone expands system knowledge, reducing retraining costs.

Transparency and Reliability
Provides evidence-based outputs, suitable for compliance and internal control environments.

mproved Accuracy
By searching the latest legal texts and internal regulations, it supplements the model’s knowledge, minimizing misinformation.

RoBERTa

High Accuracy
Trained on large-scale datasets, achieving around 98% accuracy for personal information identification and context-based analysis.

Learning-Based Improvements
Model performance improves continuously through incremental learning, and fine-tuning can be applied to specific domains.

Learning-Based Adaptability
Handles new data effectively through diverse text patterns and pre-trained learning, even when existing rules cannot process it.

FOTYPro

Internal Partner that protects Corporate Trust

Automatically managing complex regulations and personal information risks,

reducing costs and increasing trust

Key Features

LLM-based Document

Analysis

Personal Information &
Leakage Prevention

Efficient

Utilization

Applying corporate-tailored personal

information protection through

context-based analysis, not keyword-based

Handling both structured and unstructed data

and applying to varius industries

(e.g., law, finance, insurance)

Detection of sensitive and

confidential data in internal corporate

documents and DB

Prevention of unauthorized attempts

to leak transaction data,

customer consent, etc

On-premise/Cloud-based distribution is available

Providing an intuitive web-based

UI/UX for search results, log tracking, etc

Providing masking exception process and

function to print after saving masked documents

Key Technologies

MAS (Multi-Agent System)

Distributed Autonomy
Each agent makes independent decisions,

allowing the system to continue

operating partially even

in case of failures.

Collaborative Problem-Solving
Multiple agents can exchange

information and divide roles,

enabling distributed handling

of complex internal

control and risk management tasks.

Scalability
New functions or modules

can be easily integrated by attaching them

to agent units, making expansion flexible.

RAG AI (Retrieval-Augmented Generation)

Efficient Maintainability
Adding new data

to the index alone

expands system knowledge,

reducing retaining costs.

Transparency and Reliability
Provides evidence-based outputs,

suitable for compliance and

internal control environments.

Improved Accuracy
By searching the latest legal texts and

internal regulations,

it supplements the model’s knowledge,

minimizing misinformation.

RoBERTa

High Accuracy
Trained on large-scale datasets,

achieving around 98% accuracy for personal information identification and context-based analysis.

Learning-Based Improvements
Model performance improves

continuously through incremental

learning, and fine-tuning

can be applied to specific domains.

Learning-Based Adaptability
Handles new data effectively through

diverse text patterns and

pre-trained learning, even when existing rules cannot process it.

Expected Effects

Pre-blocking personal information leak

and strenghthening internal control

Maximization of operational efficiency

through corporate customization

Improvement of security through

AI-based unstructured data protection

Action to regulations related

to personal information