DTSData Enablement for AI

Short on cases? Prepare them with DTS.

DTS fills gaps in rare-case coverage, corrects skewed distributions and prepares hard-to-use data for a defined AI task.

Available onAWS MarketplaceNaver Cloud Marketplace

01The problem

The data usually exists. It just isn’t ready for the AI task.

  • Too few examples

    Rare events, fraud, and edge cases are too scarce to train or validate on.

  • Can’t be used as it is

    Regulated or sensitive records can’t go into training as they are.

  • Not in shape for the task

    Bias and skewed segments distort what the model learns.

  • Hard to check what was generated

    Teams can’t tell how closely new data follows the original, or what risk it carries.

02Data types

Start from the data you have. DTS prepares candidates.

Two kinds of data: tables and images.

Purpose

Train an income prediction model

What DTS reflects

  • Distribution of each column
  • Relationships between columns
  • Every column, including text

What you get

  • Synthetic dataCSV
  • Quality evaluation reportPDF

Purpose

Training images for a pneumonia reading model

What DTS reflects

  • Visual characteristics of the originals
  • The image count and resolution you set

The image domain is agreed during adoption.

What you get

  • Synthetic imagesZIP
  • Quality evaluation reportPDF
  • AugmentFill in scarce cases to expand the data.
  • RebalanceAdjust skewed distributions.
  • GenerateNew records to use when the original can’t be used.

03How it runs

Four steps, start to finish. The originals stay where they are.

Deployment is set to fit your environment.

  1. 01

    Add your data

    State the purpose and pick the originals.

  2. 02

    Diagnose

    Tables are profiled and their columns classified automatically. Images skip this step.

  3. 03

    Set generation conditions

    Set how much to generate and review the column classification.

  4. 04

    Review the result

    Check the quality evaluation, then download the data and the report.

IncludesAutomatic column classificationStatistical profilingPersonal data classification reviewRun statusResult report

04Quality

Generated data, checked from three angles.

Each result is scored for similarity, diversity and safety, and delivered in a result report.

Privacy-protection techniques can be considered during generation, and applying them can lower generation performance.

Quality evaluation

  • Similarity

    How closely it follows the original

    TVD2D CorrelationBERTScoreFID

  • Diversity

    How varied it is

    EntropyDuplication RateLPIPS

  • Safety

    How much risk of exposing original records

    Identification RiskLinkage RiskInference RiskCopy Risk

Metrics differ by data type.

05Deployment

Adopt DTS on its own. Inside Syntitan, it is the preparation engine.

On its own

DTS

DTS alone prepares the data an AI task is short of.

  • Expand the volume of data
  • An alternative when the original can’t be used or shared
  • Synthetic data with a quality evaluation report

Inside Syntitan

DTS + Syntitan

For now, synthetic data generation comes only with DTS on its own.

  • Syntitan: diagnose and improve data
  • DTS: generate synthetic data and evaluate its quality
  • The scope of using both together is confirmed during adoption

Customer cases

  • Finance · IBK Industrial BankFraud and transaction patterns expanded for training
  • Finance · Kyobo Life InsuranceChurn prediction data prepared with DTS

06Comparison

Other approaches work on what you have. DTS adds new records.

What happens to the rare cases with each approach.

  • Masking · anonymization

    Hides or removes sensitive values in the records you already have.

    Rare cases stay as few

  • Data sampling

    Picks a subset of the records you already have.

    Rare cases stay as few, or fewer

  • Manual labeling

    People add or correct labels on existing records.

    Labels improve, cases stay as few

  • DTS

    Adds new records that follow the original data.

    New candidate records added

07Proof

Customers, certification, and recognition

Finance · Public sector · Telecom

  • SK Telecom
  • Kyobo Life Insurance
  • ROK Army
  • ROK Air Force
  • IBK Industrial Bank
  • Woori Bank
  • Korea Heritage Service
  • NH NongHyup Bank
  • Danal
  • National Health Insurance Service
  • Kookmin University
  • 12Customers and PoCsFinance, public sector and more
  • GS 1Good Software Grade 1DTS v1.0
  • 3 yrsInnovative ProductKorea Public Procurement Service
  • 2Cloud marketplacesAWS · Naver Cloud

GS certification applies to DTS v1.0 (January 2025). The Innovative Product designation runs three years from September 2026.

  • Intellyx Digital Innovator Award 2026
  • NextRise Global Innovator 2024
  • Information Security Product Innovation Award 2024, DTS
  • GS Certified Grade 1, DTS 2025
  • Startup World Cup Finalist 2025
  • ISO/IEC 27001:2022 Information Security
  • ISO/IEC 42001:2023 AI Management
  • Emerging AI+X Top 100 2026 (AIIA)
  • AI Medical Innovation Award, AI EXPO KOREA 2025

Gartner® research

  • Deploy Domain-Specific Agentic AI to Lower AI Costs and Drive InnovationOctober 2026
  • Emerging Tech: Provider Differentiation Strategy Trends for Hyper-Synthetic DataJune 2025 · Named a Representative Vendor

Listed on Korea PPS Innovation Marketplace AWS Marketplace Naver Cloud Marketplace

08FAQ

Frequently asked questions

What is DTS?

DTS is CUBIG's Data Enablement for AI product. It fills missing cases, adjusts skewed distributions, and prepares enterprise data that is hard to use as it is for a defined AI task. Its core method is generating synthetic data that reflects the structure and characteristics of the original, and it evaluates each result for similarity, diversity and safety. Deployment is set to fit the customer environment, and DTS is listed on AWS Marketplace, Naver Cloud Marketplace and the Korea PPS Innovation Marketplace.

How does DTS handle personal data?

DTS classifies personal data columns in the original so teams can review and correct them, and it measures identification, linkage and inference risk in the generated result as safety metrics. Privacy-protection techniques such as differential privacy can be considered during generation, and applying them can lower generation performance. Synthetic data alone does not guarantee complete protection of personal data, so review is needed before use.

Can DTS run without Syntitan?

Yes. DTS can be adopted on its own. Synthetic data generation is currently provided through on-premises DTS and is not yet part of the Syntitan SaaS.

What data problems does DTS solve?

Four kinds: too few examples of rare cases, data that cannot be used as it is, data skewed in one direction, and no way to check the quality of newly generated data.

Can DTS work without moving the original data?

Yes. DTS can be deployed to run inside the customer environment, and it does not change the content of the original dataset. It returns the generated data with a quality evaluation report, so it can also fit environments where data is hard to move out, such as isolated networks.

How is DTS different from Syntitan?

DTS is CUBIG's Data Enablement for AI product, and it can be adopted on its own. Syntitan is the AI-Ready Data Platform: it assesses data readiness, validates data fit for a use case under the same model and evaluation conditions, and keeps each version with the conditions it was checked under. Inside Syntitan, DTS works as the data preparation and supplementation engine. Synthetic data generation is currently provided through on-premises DTS.

Is CUBIG's data optimization the same as DTS Data Enablement?

No. Data optimization is CUBIG's overall operating approach: preparing and validating data for a specific AI task, and carrying the evidence behind decisions through real use and later changes. DTS is the product within it that fills missing cases, adjusts skewed distributions, and handles restrictions on how data can be used.

Does Data Enablement mean DTS also provides a data catalog or governance?

Data Enablement is used broadly in the market. When CUBIG describes DTS, it states the product scope alongside it: DTS fills missing cases and adjusts skewed distributions so data fits a defined AI task.

Prepare your data for the AI task it’s missing.

Start with one AI task and the cases it doesn’t have enough of.

Available onAWS MarketplaceNaver Cloud Marketplace