Pulse

Pulse, from Pulse Software, Corp., is a document intelligence platform that uses OCR, layout, and vision models to produce structured outputs from complex documents for enterprise and AI-native teams. It is offered through an API and web platform, with private-cloud and on-prem deployment options for enterprise customers.

What It Is

Pulse converts unstructured documents such as financial statements, medical records, claims packets, contracts, and rent rolls into structured, confidence-scored data. The API is described as zero-shot and document-type agnostic. The company's blog describes outputs such as cell-level bounding boxes and confidence scores, plus page classification and file splitting before schema-based extraction.

How the Document Vision Pipeline Works

The homepage describes a five-step pipeline: layout understanding with component detection models, low-latency OCR for individual components, reading-order algorithms for different document types, table structure recognition and parsing, and fine-tuned vision-language models for converting charts, tables, and figures. Layout is analyzed before content is read, and multi-column or irregular layouts are processed in a human-like reading sequence.

Deployment and Security

Enterprise customers can run Pulse in a private VPC or on-prem environment, using Docker, Kubernetes, or internal API gateways, with logging and metrics over document type and latency. The vendor lists SOC 2 Type II, ISO 27001, GDPR alignment, and a HIPAA BAA for enterprise plans. Its blog also describes a Snowflake workflow and a Claude connector for passing document URLs and receiving structured data.

Use Cases

Pulse publishes industry pages for finance, healthcare, insurance, legal, real estate, and supply chain. The vendor states it has processed over 1 billion pages.