Sun Finance Automates Identity Verification with AWS Generative AI

Introduction

Identity verification (IDV) has long been a critical component in the fintech sector and serves as the first line of defense against fraud. Traditional solutions often suffer from insufficient accuracy, lengthy processing times, and high costs. Recently, global consumer fintech company Sun Finance unveiled its new identity verification pipeline built on AWS generative AI technologies, achieving breakthrough improvements across three dimensions: accuracy, cost, and efficiency.

Core Solution: The OCR + LLM Combination

The core philosophy behind Sun Finance's solution is the deep integration of specialized OCR tools with the structured reasoning capabilities of large language models (LLMs), rather than relying solely on any single technology.

Specifically, the solution integrates three major AWS AI services:

• Amazon Textract: Responsible for high-precision optical character recognition (OCR) on identity documents, extracting text and structured field information.
• Amazon Bedrock: Invokes large language models to perform intelligent structuring, field validation, and error correction on the raw data extracted by Textract, compensating for the limitations of traditional OCR in complex layouts and multilingual scenarios.
• Amazon Rekognition: Used for facial comparison and liveness detection for visual-layer fraud identification.

Sun Finance's experience demonstrates that using OCR alone or LLMs alone both have limitations, but when the two are chained together, the system's document information extraction accuracy jumped from 79.7% to 90.8% — an improvement of over 11 percentage points.

Remarkable Efficiency and Cost Advantages

Beyond accuracy improvements, the solution's operational efficiency gains are equally impressive:

• Processing time: Reduced from a manual review process that previously took up to 20 hours down to under 5 seconds for automated completion, delivering a near real-time identity verification experience.
• Per-document cost: Reduced by 91%, representing enormous operational cost savings for financial institutions that process massive volumes of documents daily.

These results are attributable to Sun Finance's adoption of a serverless architecture design. The entire AI identity verification pipeline is built on AWS serverless services, eliminating the need to manage underlying infrastructure. It automatically scales elastically based on business volume, ensuring stability under high-concurrency scenarios while avoiding idle resource waste.

Industry Analysis: Generative AI Is Reshaping Financial Risk Management

Sun Finance's case reflects a clear industry trend — generative AI is evolving from an experimental "nice-to-have" tool into a core component of financial risk management infrastructure.

Traditional identity verification solutions rely heavily on manual review and rules engines, often struggling to keep pace with the diverse global document formats and sophisticated forgery techniques. The introduction of generative AI brings two major transformations:

1. A leap in comprehension capabilities: LLMs can understand the contextual semantics of documents, perform intelligent error correction and logical validation on OCR results, and significantly reduce misextraction and missed extraction.
2. A paradigm shift in architecture: With cloud-native serverless architecture, enterprises can rapidly deploy and iterate verification pipelines without investing heavily in building their own AI infrastructure.

Notably, Sun Finance chose a hybrid approach of "specialized tools + general-purpose models" rather than relying entirely on a single large model to handle all tasks. This pragmatic strategy is gaining increasing recognition across the industry — let specialized tools do what they do best, and let LLMs handle integration and reasoning.

Outlook

As global regulatory requirements for KYC (Know Your Customer) and anti-money laundering compliance continue to tighten, demand from financial institutions for efficient, low-cost, and highly accurate identity verification solutions will only continue to grow. Sun Finance's implementation provides an invaluable reference for the industry: by thoughtfully orchestrating multiple AI capabilities, enterprises can achieve breakthroughs in both cost and efficiency without sacrificing accuracy.

Looking ahead, as multimodal large model capabilities continue to evolve and cloud providers like AWS continue to enrich their AI service portfolios, similar intelligent verification pipelines are poised to expand into more industries and use cases, becoming a vital cornerstone of the digital trust ecosystem.