Case Study

Accelerating Insurance AI Development with Privacy-Safe Synthetic Data

Customer: Sixfold (Insurtech – AI for Insurance Underwriting)
Solution: DataFramer – Synthetic Data Generation Platform

---

Background

Sixfold is an insurtech building AI systems that help insurers underwrite **Life & Health (including Life and Disability)**as well as Property & Casualty (P&C) products. Its platform ingests complex insurance data and produces structured analysis and summaries that enable underwriters to make faster, more consistent, and more compliant decisions.

As Sixfold scaled its platform across insurance lines and enterprise customers, access to high-quality data became a limiting factor. Much of the data Sixfold works with is highly sensitive, tightly regulated, and customer-owned. Relying on real production data slowed development, complicated evaluations, and restricted who inside the company could safely interact with realistic underwriting scenarios.

---

The Challenge

Sixfold’s AI systems operate on data that includes medical records, insurance applications, commercial submissions, and third-party research. This data often contains PII and PHI, making it difficult to reuse across teams, share with customers during pilots, or rely on consistently for testing and evaluation.

At the same time, Sixfold needed stable and repeatable datasets to evaluate model quality. LLM and workflow evaluations varied from run to run, rare but high-impact underwriting scenarios were under-represented, and regression testing became unreliable as real data changed.

Product and design teams faced a parallel issue. Without access to realistic underwriting data, they struggled to review AI outputs, validate user experiences, and align on what “good” looked like before features shipped.

---

The Solution

Sixfold adopted DataFramer to generate high-fidelity synthetic underwriting datasets that preserve the statistical structure of real insurance data while removing all real-world identifiers.

Using DataFramer, Sixfold could instantly create datasets that reflected real distributions across Life & Health and P&C underwriting. Medical conditions, lifestyle attributes, business profiles, and risk signals appeared in realistic combinations, while rare and edge-case scenarios could be deliberately emphasized for testing and evaluation.

Each dataset came with built-in quality and privacy assurances, giving Sixfold confidence that synthetic data could be used safely across engineering, product, customer demos, and vendor evaluations.

---

How Sixfold Uses DataFramer

For engineering and data science teams, DataFramer’s API became a core part of the development workflow. Teams could generate repeatable synthetic datasets on demand for model training, LLM and RAG evaluation, stress testing, and regression testing. These stable “golden datasets” made it possible to compare model versions, measure improvements reliably, and share data safely with external partners without exposing customer information.

For product leadership and design teams, DataFramer’s UI provided a way to explore realistic underwriting scenarios without privacy risk. Teams could review AI-generated summaries, test edge cases, and validate user experiences using data that looked and behaved like production, but was safe to share broadly. This enabled earlier alignment and reduced downstream rework.

---

Results

Medical Professional Validation: Before gaining confidence and signing up, Sixfold had DataFramer’s Life & Health datasets reviewed by medical professionals, who reviewed them as perfect. Additionally, DataFramer had the datasets reviewed with MDs (Medical Doctors), who reviewed them as meticulous. This validation gave Sixfold the confidence to adopt DataFramer and later expand into P&C lines.

By introducing synthetic data as a standard layer in its AI platform, Sixfold significantly accelerated development and evaluation cycles. Enterprise customer pilots and demos were no longer blocked by data-sharing concerns, and AI quality could be measured consistently across use cases and insurance lines.

Just as importantly, Sixfold reduced its reliance on production customer data, strengthening its trust posture with insurers while maintaining speed and flexibility internally.

---

Why DataFramer

Sixfold chose DataFramer for its ability to produce distribution-aware synthetic data with strong privacy guarantees, delivered through both an API for engineering teams and an intuitive UI for product and leadership stakeholders. This combination allowed synthetic data to scale beyond experimentation and become a foundational capability across the organization.

---

Summary

With DataFramer, Sixfold transformed synthetic data from a compliance workaround into a strategic enabler. The platform now supports faster, safer, and more scalable AI development across Life & Health and P&C underwriting, while preserving the trust of enterprise insurance customers.