headless is a financial data API service that aggregates, normalizes, and classifies financial data from over 78 Korean financial institutions, including banks, credit card companies, and the National Tax Service (Hometax). It aims to simplify access to diverse financial data streams by providing a single, unified API endpoint, abstracting away the complexities and differences inherent in connecting to multiple disparate financial systems.

The core methodology of headless revolves around three key pillars:

1. Unified Endpoint and Abstraction: headless integrates various authentication methods and data retrieval mechanisms from 78+ financial institutions into a single RESTful API endpoint, typically https://api.h6s.ai/v1/data-jobs. This architecture eliminates the need for clients to manage institution-specific authentication flows, data formats, or API variations. The service acts as an intermediary, handling all communication, authentication (e.g., certificates, passwords), and interface changes with the underlying institutions. For instance, a single POST request to /api/v1/data-jobs can initiate a data collection job across different banks using a common connector_id and action (e.g., bank.transactions).

2. Standardized Schema and Normalization: Data retrieved from various financial institutions, often in inconsistent or proprietary formats, is transformed and normalized into one of 12 predefined standard schemas. This ensures that regardless of the source institution, the response structure for a given data type (e.g., bank transactions) remains consistent. For example, differing merchant names and transaction categories are mapped to standardized fields like category (e.g., "F&B/카페") and amount. This process involves:
   • AI Classification: Unstructured or semi-structured data, such as merchant names, is automatically processed and classified into standardized categories using AI algorithms. This helps in structuring raw, heterogeneous data into a clean, queryable format.

3. Robust Data Collection and Operational Management: headless provides automated handling for common challenges in financial data retrieval:
   • Resilience: It automatically manages network errors, timeouts, and retries. If an institutional interface changes, headless adapts automatically, minimizing operational burden for the client.
   • Scalability: For large data queries, the system supports automatic chunking (pagination), ensuring efficient data transfer without requiring client-side implementation of such logic.
   • Asynchronous Processing: The API supports various integration patterns, including synchronous HTTP for immediate responses, asynchronous HTTP with polling, webhooks for callback notifications upon job completion, and message queue integration (e.g., SQS, Kafka) for event-driven architectures. This flexibility allows clients to choose the integration model best suited to their system architecture.

The typical workflow for a user involves three steps:

4. Connect: Register financial institution certificates and configure connectors via a POST /api/v1/connectors request, providing providerId and credentials.
5. Collect: Initiate data collection jobs using a simple POST /api/v1/data-jobs request, specifying connectorId and the desired action (e.g., bank.transactions). The system handles chunking and retries automatically.
6. Normalize: Retrieve the processed and normalized data via a GET /api/v1/data-jobs/{job_id}/results request. The response adheres to a standard schema, abstracting institutional variations.

For security, headless implements:

• AES-256 Encryption: All sensitive data, including certificates, passwords, and access information, is encrypted using AES-256 and stored in isolated, secure storage.
• Isolated Storage: Credentials and sensitive information are managed in isolated, hardened storage environments, preventing direct exposure on client servers.
• Access Control: API key-based access control allows fine-grained permission and usage rate management.
• Audit Trails: Access logs are maintained for auditing purposes.
• Environment Separation: Development (Sandbox) and production (Live) environments are completely separated.

The service boasts a developer-friendly experience with SDKs available for popular languages (Node.js, Python, Java, Kotlin, Go) and curl examples, enabling quick integration and first API calls within minutes. Its pricing model includes a free tier with limited API calls and an Enterprise tier offering unlimited calls, SLA guarantees, and dedicated support.