Skip to content
Haystac — Document Parsing Service for IDP
VENDORS 3 min read

Haystac — Document Parsing Service for IDP

US-based IDP startup offering a patent-pending Document Parsing Service that extracts data from structured, semi-structured, and unstructured documents using trainable local language models.

Overview

Haystac is a Boston-area intelligent document processing company built around a single product: the Document Parsing Service (DPS). The company was founded by Barak, who previously led Solaris Development Inc. from 2003 into a recognized healthcare contract software development firm serving Partners Healthcare, MGH, BWH, and Dana Farber Boston Children's Hospital. CTO Eli brings over 30 years of experience in distributed systems, rules engines, and computational linguistics, including co-founding Sapiens and directing development of automated document classification and retention management products.

The advisory board adds significant IDP industry depth. Anthony Macciola, currently Chief Innovation Officer at ABBYY, holds over 45 patents in mobility, text analytics, and image processing and previously served as CTO at Kofax, where he directed the company's move into mobile capture and natural language processing. Trevor Lund, former CEO of Metalogix GmbH, and Tom Reilly, former CTO at StoredIQ through its IBM acquisition, round out the technical advisory bench. The combination of a healthcare software pedigree in leadership and enterprise IDP expertise on the advisory board shapes Haystac's positioning as a practical, deployment-focused alternative to the larger platforms.

No funding rounds, partnerships, or product launches beyond the core DPS offering were identified in the research window. The company's public profile remains limited, which is consistent with an early-stage, bootstrapped posture.

How Haystac Processes Documents

DPS is built on what Haystac describes as a trainable local language model framework, distinct from cloud-hosted large language models. The core claim is that this architecture eliminates the upfront configuration work that burdens traditional OCR and template-based extraction platforms. Deployments that previously required weeks or months of setup are, according to the vendor, achievable in hours or days.

The service handles three document categories: structured forms such as W-2s, tax returns, IDs, and insurance claims; semi-structured documents including invoices, bills of lading, receipts, and explanations of benefit; and unstructured documents such as commercial leases and purchase contracts. This breadth positions DPS as a general-purpose extraction layer rather than a single-vertical tool, though no independent benchmark results or third-party accuracy figures have been published to substantiate the coverage claims.

The "zero-setup initiative" is the product's stated differentiator. Where most classification and extraction platforms require document samples, labeling, and model training before processing begins, Haystac claims DPS can integrate and deploy without that upfront investment. The patent-pending status of the underlying AI technology suggests the approach is novel enough to pursue IP protection, but the claims remain vendor-reported and unverified.

Use Cases

Healthcare and Insurance

The founding team's background in healthcare contract software development at institutions including MGH and Dana Farber points to a natural early market. DPS handles explanation of benefit documents, insurance claims, and structured forms such as W-2s and tax returns. The semi-structured document capability is relevant for payers and providers processing high volumes of EOBs and remittance documents where layout varies across counterparties.

Commercial leases, purchase contracts, and tax documents represent the unstructured document category where template-based extraction typically fails. DPS targets this gap directly. For organizations processing varied contract formats across counterparties, a zero-setup approach reduces the per-document-type configuration cost that makes traditional platforms expensive to scale. This use case overlaps with the broader accounts payable automation and contract intelligence space occupied by vendors such as Eigen Technologies and Kira Systems.

Technical Specifications

Feature Specification
Document Types Structured forms (W-2, tax return, ID, claim), semi-structured (invoice, bill of lading, receipt, EOB), unstructured (commercial lease, purchase contract)
Input Formats Not disclosed
Output Formats Not disclosed
Processing Pipeline Trainable local language model framework; patent-pending AI
API/Integration Not disclosed
Deployment Options Not disclosed
Certifications None disclosed
Claimed Accuracy Not disclosed; vendor claims zero-setup deployment in hours vs. weeks

No independent benchmark results, accuracy figures, or third-party analyst coverage have been identified. All performance claims are vendor-reported and unverified.

Resources

Company Information

Boston, MA, United States. Founding year not publicly disclosed. Led by Barak (CEO) and Eli (CTO), with an advisory board that includes Anthony Macciola of ABBYY, Trevor Lund, Tom Reilly, Peggy Gartin of Assist Legal Technologies, and Christopher Moran of Shell. No external funding rounds disclosed.