One platform. Four copilots. Your documents.
Retrieval, document intelligence, agents, and dashboards — built as one system, deployed inside your infrastructure, and accountable to your sources.
Every answer shows its work.
Mynd Copilot answers from your manuals, SMS, circulars, and regulations — and every answer carries the document, page, and paragraph it came from.
- Thousands of documents, one question box
- Photo and diagram understanding for troubleshooting
- Answers in the terminology your crews actually use
- History-aware: "what did we do last time this alarm came up"
The main engine shows a scavenge fire alarm. What's the immediate action?
Reduce to slow speed and stop fuel to the affected unit, then increase cylinder lubrication and confirm the scavenge space fire has not spread 1Maker §7.4.
Engine room emergencies
Maker manual §7.4 · p.146
Grounded, not generative guesswork.
When someone asks a question, the platform retrieves the relevant passages from your approved corpus first, then writes the answer from those passages alone. The sources appear beside the answer as citations your team can open in one click.
If it can't cite, it says so.
A question your documents don't answer gets "not found in your sources" — not a confident invention. In work where a wrong answer becomes a finding, a detention, or an incident, this is the property that matters most.
Built for the 3 a.m. question.
An engineer mid-breakdown, a mate preparing for port state control, a new joiner learning your procedures — they ask in plain language, attach a photo if it helps, and get the procedure back with the maker's manual page cited.
Trained on your corpus, current by design.
Upload a revised manual and the old revision is superseded, not blended. The copilot answers from the controlled versions your DPA approved.
Documents that verify themselves.
Mynd Docs reads, classifies, and validates the paperwork your operation runs on — crew certificates, COCs, STCW endorsements, medicals, invoices, POs, KYC — and flags problems before they reach an approver's desk.
- Certificate, invoice, and KYC classification out of the box
- Expiry and mismatch alerts with stated reasons
- Search and retrieve any processed record in seconds
- Human review queue for everything the system isn't sure about
Master · Issued IN · valid for vessel type
Reads like a document checker, at machine speed.
OCR and language models extract every field — names, numbers, issue and expiry dates, issuing authorities — from scans, photos, and PDFs, including the imperfect ones that arrive from ships and agents.
Checks eligibility, not just legibility.
Extracted data is validated against your rules: certificate valid for the rank and vessel type, medical in date at sign-on, endorsement matching the flag. Expired, mismatched, or incomplete records are flagged with the reason stated.
The crew-to-ship matrix, always current.
Manning status across every vessel at a glance — who signs on, who signs off, whose documents block them, and advance alerts well before contract expiry dates become port problems.
Beyond crewing.
The same engine runs three-way matching of POs, goods receipts, and invoices, and screens vendor documents — with anomalies flagged for a human decision.
Workflows that run themselves. People stay in charge.
Mynd Flow chains the platform's intelligence into governed workflows — approvals, crewing processes, onboarding, email triage — with humans approving what matters and automation carrying the rest.
- Multi-level approval flows with automatic escalation
- Candidate screening mapped to rank and vessel requirements
- Email thread summaries and drafted replies
- Full action log for every automated step
Purchase request
raised
Document checks
Mynd Docs
Human approval
role-based
Routed and logged
complete
Approvals without the chase.
Purchase requests, contracts, and internal approvals route themselves, escalate when they stall, and leave a complete trail — status visible to everyone, email dependency gone.
Crewing and HR, end to end.
Applications screened against vessel-specific requirements, credentials verified, interview scheduling and onboarding checks handled by an assistant — your recruiters decide, the platform does the paperwork.
An inbox that triages itself.
Incoming vendor and port mail categorized, prioritized, and summarized, with action items extracted and suggested replies drafted for the routine traffic. Your team answers what needs judgment.
Governed by design.
Every agent action is permissioned, logged, and reversible. Automation runs inside the same role-based access and audit trail as everything else on the platform.
Your systems stay the record. We remove the manual work between them.
The platform is API-first: it reads from and writes to SAP, DANAOS, AFSYS, HRMS, and your document stores through secure connectors — no rip-and-replace, no parallel database of truth.
- SAP, DANAOS, AFSYS, HRMS via secure APIs
- File shares, SharePoint, and mail ingestion
- MCP-based connector architecture
- Scoped credentials and per-connector audit logs
Built to sit on top.
ERP, crewing, and planned-maintenance systems hold the records; the platform automates the reading, checking, answering, and routing your people do around them today.
Connectors, the modern way.
Integrations are built on MCP — an open standard for connecting AI systems to enterprise tools — so new connections are additions, not projects. File shares, mail, and databases connect the same way.
Your IT team keeps control.
Every connector is scoped, credentialed, and logged. Integration runs through interfaces your team approves — nothing crawls what it shouldn't.
The fleet, on one screen. Ask it anything.
Noon reports, voyage reports, bunker logs, and maintenance records are extracted automatically and consolidated into ship and fleet dashboards you can question in plain language.
- Automatic noon-report and log extraction
- Ship and fleet dashboards with drill-down
- Plain-language querying with cited figures
- Predictive maintenance alerts from your own history
No more retyping ship reports.
Reports arrive as they always have — the platform extracts the numbers, validates them, and files them. Data entry disappears; the history becomes queryable.
Ship level and fleet level.
Per-vessel dashboards for speed, consumption, cargo status, crew compliance, and maintenance status; a consolidated fleet view for trends, benchmarking, and the KPIs leadership actually asks for.
Ask, don't filter.
"Show me all tankers with surveys due this quarter." "Which vessels' bunker consumption drifted above charter-party terms this month?" Plain-language questions, answered from live fleet data — with the underlying figures cited.
From history to foresight.
Maintenance records, running hours, and failure patterns feed predictive alerts — recurring issues surfaced by equipment type, vessel class, or maker, before they become off-hire days.
Private by architecture. Not by promise.
The platform deploys in your private cloud or entirely on-premise. Your documents, models, and indexes are exclusive to you — nothing is shared across customers, and nothing is used to train anyone else's models.
- Private cloud or on-premise
- Customer-exclusive models and indexes
- Encryption at rest and in transit
- Role-based access and full audit trails
Your perimeter
FreshMynds Platform
Models · indexes · documents
Deployment on your terms.
On your servers behind your firewall, in your private cloud tenancy, or hosted for you — your IT policy decides.
Exclusive by default.
Models, embeddings, and indexes are built for you and stored in your environment. There is no shared inference pool learning from your manuals.
Enterprise controls throughout.
Encryption in transit and at rest, role-based access for officers, office, and auditors, and a complete audit trail of every question, answer, document decision, and agent action.
Scales with the fleet.
Architecture sized from a five-vessel manager to a two-hundred-vessel expansion plan without re-engineering.
Ready when your documents are.
A demo takes an hour. A proof of concept takes weeks, on your own documents, against success criteria we agree together.
