Harbor

Harbor is a clinical trial software platform that provides electronic data capture (EDC) and eClinical data management tools for clinical studies. The platform is designed to automate the process of building case report forms (CRFs), capturing source data, and monitoring trial data, with compliance built into its core architecture. Harbor serves a range of customers including academic researchers, investigator-initiated trial teams, commercial sponsors, and contract research organizations (CROs).

The platform positions itself as an alternative to legacy EDC systems such as Medidata, Viedoc, and Veeva, with a focus on reducing study startup timelines, minimizing manual data entry, and providing real-time data visibility. Customers cited include BioDynamik and NOVA Genomics, with reported outcomes including EDC builds delivered in as few as nine days and study startup processes reduced from two months to two days.

Core Product Modules

• Magic Build: Parses a protocol PDF and automatically generates draft CRFs and source document packets. Harbor states this process takes approximately 10 minutes, replacing build timelines that can otherwise run 8–12 weeks with legacy vendors.
• Magic Capture: An AI-powered data extraction tool that processes source documents uploaded by clinical sites and transcribes data directly into the EDC. Harbor reports this reduces manual data entry time by approximately 90%, converting roughly two hours of typing into 15 minutes of review.
• Magic Monitor: A remote monitoring module that displays source documents side-by-side with EDC data. Each data point is scored for extraction confidence and clinical relevance, enabling centralized and risk-based monitoring without requiring sites to change their existing workflows.

Key Capabilities

• Automated generation of eCRFs and source documents directly from protocol PDFs
• AI-driven source data extraction to reduce transcription errors and manual entry burden
• Real-time data visibility with zero latency from the moment source documents are uploaded, eliminating the typical 7–10 day delay associated with manual site entry
• Risk-based and remote monitoring with confidence scoring on extracted data points
• Immutable audit trails, electronic signatures, and role-based access controls enforced by default
• A comprehensive Verification and Validation (V&V) package included with every license, covering risk and traceability matrices

Security and Compliance

• Compliant with 21 CFR Part 11, HIPAA, and GDPR
• Data encrypted at rest using AES-256 and in transit via TLS 1.2 or higher
• Each study is hosted in logically isolated databases to prevent cross-contamination between studies
• Infrastructure runs on multi-zone redundancy with automated daily backups and recovery capability in the event of a regional failure
• Zero data retention policy: organizational data is used only for the clinical trials being run on the platform
• Business associate agreements (BAA) enforced with all partners

Pricing and Licensing

• Academic and Non-Profit (Free): Intended for university research and investigator-initiated trials. Includes the full AI suite (Magic Build, Magic Capture, Magic Monitor), unlimited users and patients, standard email support, and standard data export.
• Commercial Study (starting at $2,000/month): Designed for growing teams. Includes everything in the Academic tier plus the 21 CFR Part 11 validation package, priority support via a dedicated Slack channel, training for sites and clinical research associates (CRAs), and a guaranteed uptime SLA.
• Enterprise / CRO (volume pricing): Targeted at CROs and sponsors managing three or more active trials. Includes everything in the Commercial tier plus a master service agreement (MSA), SSO/SAML authentication, and integrations with CTMS and eTMF systems, along with a dedicated customer success manager.

Harbor states that all automation features, including the full Magic Build, Magic Capture, and Magic Monitor suite, are included across all license tiers and are not gated behind additional paywalls.