Maritime & Shipping
US
5
4 months; ongoing
The client is a US-based ship management company providing commercial and technical management for a fleet of 40+ dry bulk and tanker vessels.
Every port call, charter, and crew rotation generates documents that move between charterers, port agents, class societies, flag states, and the client’s back office. Document control directly affects cargo release, charter payments, and inspection readiness — a single missing or expired certificate can hold a vessel in port.
The client’s document flow ran on email and manual re-keying.
Bills of lading, statements of facts, charter parties, crew certificates, class certificates, and survey reports arrived from dozens of counterparties in inconsistent formats and naming conventions. Operations staff downloaded attachments, classified them by hand, and re-entered key fields into the fleet management system. Certificate expiry dates lived in per-vessel spreadsheets.
This created several business risks:
Superintendents and operations managers were pulled into administrative work instead of managing vessels.
We designed and built an AI document processing platform that automates the document flow from the inbox to the fleet management system.
Maritime is a regulated industry, and document automation there is earned flow by flow. Rather than promise full coverage on day one, we scoped the first release around the platform core and the three flows with the highest volume and clearest business cost: bills of lading, statements of facts, and crew certificates. Each new document type is added through configuration and a validation cycle — not a new project.
For crew certificates, the same pipeline feeds one more destination: a centralized expiry register with automated alerts, replacing per-vessel spreadsheets. Because the register is a downstream output of the document flow — not a separate system to maintain — it stays current without extra data entry. The office now sees upcoming fleet-wide renewals weeks before a port call.
Modern fleet management systems can store and file documents. The bottleneck sits upstream: intake from dozens of counterparties, classification of inconsistent formats, extraction from scans, and validation against the fleet’s own register. That is the work this platform does before a document ever reaches the FMS. The integration itself is API-based and system-agnostic — the pipeline’s structured output defines the contract, not the specific FMS behind it.
The solution runs entirely within the client’s Azure environment:
All document storage, AI processing, and application logic remain within the client’s Azure environment. No document or extracted business data is transferred to external AI providers or third-party processing platforms.
The initial 4-month engagement delivered the platform core, the fleet management system integration, the certificate register, and three document flows validated in production. A smaller team now onboards the remaining document types — charter parties, class certificates, survey reports, port documents, and invoices — each through its own configuration and validation cycle.
Every processed document also lands as structured, indexed data. That archive is the base layer for retrieval across historical records — maintenance, repairs, and surveys — scoped for the next phase.
For automated flows, document handling moved from manual classification and re-keying to automated extraction and structured filing. Processing time per document dropped from ~30 minutes to under 3, accelerating cargo documentation and charter payments.
Operations staff and superintendents handle a growing volume of documents without additional headcount, and the share of their time spent on intake and data entry continues to shrink as each new flow goes live.
The centralized certificate register and automated expiry alerts replaced spreadsheet tracking for automated certificate types. Renewals surface weeks before port calls, rather than during pre-inspection scrambles.
Every automated document enters the system with consistent metadata and a traceable processing path. Validation rules and confidence flagging cut re-keying errors, and the structured log provides clear audit and inspection evidence.
Tell us what you’re building or where you’re stuck. We work with engineering and product teams on custom software, AI & ML, cloud infrastructure, DevOps, and UI/UX — from early scoping to long-term delivery. One conversation is usually enough to know whether we’re the right fit.
Email: [email protected]
We and our partners use technology such as cookies on our site to personalize content and ads, provide social media features, and analyze our traffic. Click “Accept” to consent to the use of this technology across the web.
