The complete guide to ADRSA-MARINE.

Everything the product and demo team needs to understand the platform — the concepts, the roles, the risk model, and the AI that powers it. No engineering knowledge required.

02Core concepts

Five ideas explain the whole platform. Everything else builds on these.

03Roles & permissions

Three kinds of actor use the system. A company can have many vessels and many team members; each team member is either an Admin (editor) or a Standard user (viewer).

Report permission matrix

04Setup in 5 steps

The path a brand-new company takes from sign-up to its first risk report.

1. Onboard the company
   Register the operator: company name, registration number, email, location, logo, and the primary “Master User” login. This is the account everything else hangs off.
2. Configure the risk matrix
   Choose the matrix size and define the severity and likelihood levels and the Low/Medium/High thresholds. Save a draft, then generate the final matrix. Details →
3. Add vessels
   Upload each vessel’s questionnaire (Q88 / OVPQ / VPQ) to have its particulars read automatically, or enter them manually. Details →
4. Invite & approve the team
   Team members self-register and pick your company; you approve them and set each as Admin or Standard. Details →
5. Build risk reports
   Pick a vessel, describe a job, and let the AI generate the steps, hazards, consequences, and mitigations — then refine, score, and save. Details →

05Risk matrix

The risk matrix is how a company decides what “risky” means. It is configured once, then applied to every consequence in every report.

The scoring model

Each risk score is banded into Low, Medium, or High using thresholds the company sets. Example: an outcome that is Major (severity 4) and Likely (likelihood 4) scores 16 — High, red.

What the company configures

• Matrix size — 4×4, 5×5, or 8×8, depending on how granular the operator wants to be.
• Severity levels — named steps for how bad an outcome is (e.g. Negligible → Catastrophic), mapped to functional consequence groups.
• Likelihood levels — named steps for how probable an outcome is (e.g. Rare → Almost Certain).
• Probability thresholds — the score ranges (start–end) and colours that define Low / Medium / High tiers.

The configuration can be saved as a draft and then generated into the final matrix. Once generated, edits keep it generated and the matrix stays in sync.

Final risk matrix review

A colour-coded grid — green (low), yellow (medium), red (high) — showing the score for every severity × likelihood combination, alongside the probability thresholds. This is the operator’s risk model at a glance, and the same bands appear on every report’s Score Reference panel.

06Vessel inventory

A company’s fleet. Each vessel is identified by its IMO number or call sign (at least one is required), and a vessel name. The platform prevents duplicates — if a vessel with the same name, IMO, or call sign already exists, it tells you to go add a job instead.

Two ways to add a vessel

What’s captured for each vessel

Vessel data is organised into two sections, matching the screens:

Fields the source document doesn’t contain come back blank for the user to complete. Ship type is chosen from a managed list of vessel types, each with its own sub-types (e.g. Tankers → VLCC, Suezmax, Aframax).

07Risk reports

A risk report is the heart of the platform: a structured, scored assessment for one job on one vessel. Every report is built through a guided four-step wizard.

The report structure

A report has multiple job steps; each step has hazards; each hazard has consequences; each consequence has mitigation measures. Every consequence carries a score.

The four-step wizard

1. Job Details
   Select the vessel and describe the job (name + description). The AI can suggest the job steps, or you add them yourself.
2. Hazards
   For each job step, list the potential hazards. AI suggestions are available per step; cards can be reordered by drag-and-drop.
3. Consequences
   For each hazard, add its consequences and set the severity and likelihood — the platform computes the Initial score and its Low/Med/High band automatically.
4. Mitigations
   For each consequence, add mitigation measures (AI-suggested or manual). An optional Residual score captures the risk that remains after controls are applied.

Throughout, a Score Reference panel shows the company’s Low/Med/High ranges and a Step Summary tracks counts (hazards, consequences, how many have mitigation). Reports can be saved as draft or marked completed.

Initial vs. residual risk

Every report gets a unique Task ID

On creation, each report receives a human-readable ID: the first four letters of the vessel’s name (zero-padded if shorter) + a six-digit number — e.g. ivor432657 for “Ivory Carrier”, or 00fg452278 for “Fg Container”. It’s unique and stays fixed for the life of the report.

Finding & downloading reports

• Search a previously generated report by report name, IMO number, call sign, or vessel name.
• Filter the report list by recency (e.g. last 7 / 30 / 90 days), with paging for large fleets.
• Download a finished report as an encrypted PDF — available to every role, including Standard users.

08The AI engine

The AI is what makes ADRSA-MARINE fast — and what makes it trustworthy. It doesn’t guess; it reads.

Grounded in real regulation

Behind the platform is a Knowledge Base — a library of maritime regulatory documents (IMO conventions, IMDG code, class and flag guidance, and more). When you ask the AI for hazards or mitigations, it retrieves the most relevant passages from that library and answers from them. Two safeguards keep it honest:

• Sourced. Every suggestion is drawn from actual documents. Internally each is traceable to its source.
• Validated. If the sources don’t support an answer, the AI says so rather than inventing — and the platform filters those “not enough information” responses out before they ever reach a report.

Where the AI helps

Hazards are drawn from a structured Marine Industry Hazard Wheel of 14 categories — Physical, Chemical, Biological, Environmental, Mechanical/Equipment, Electrical, Fire/Explosion, Ergonomic, Operational/Process, Human Factors, Organizational/Management, Security, Environmental Impact, and Navigational — so results stay consistent and marine-specific.

09Team management

Companies control who has access and what they can do.

1. A team member self-registers
   They sign up, choose your company from a list, and set an email + password. Their account starts as Pending.
2. You review the request
   On the User Management screen you see everyone, with counts of total users and pending approvals.
3. You decide
   Approve or Reject, set the role to Admin or Standard, and later Activate, Deactivate, or remove accounts as needed.

Only active accounts can sign in. Pending, rejected, and deactivated users are turned away with a clear message.

10Security & access

• Tenant isolation. Every request is scoped to the signed-in company — one operator can never see another’s vessels, reports, or team.
• Role-based access. Editing is gated by role; viewers can’t change data, and members only touch their own company.
• Secure sessions. Sign-in issues a short-lived access token that the app quietly renews while you’re active, so sessions stay alive without long-lived credentials sitting around.
• Self-service password reset. Enter your email, receive a 4-digit code (valid ten minutes), verify it, and set a new password. Logged-in users can also change their password from Settings.
• Encrypted report PDFs. Downloaded reports are password-protected.
• Trusted HTTPS. The platform is served over encrypted, certificate-verified connections.

11Demo walkthrough

A crisp ten-minute story that shows the whole platform.

1. Onboard a company
   Show the onboarding form — “this creates the operator’s HQ account.”
2. Configure & generate the risk matrix
   Pick a 4×4 matrix, set severity/likelihood, generate it, then open Final Risk Review for the colour grid. “This is how this company defines risk.”
3. Add a vessel automatically
   Upload a sample questionnaire; point out the particulars were read for you.
4. Register & approve a team member
   Sign up as a new user (show “Pending”), then approve them as an Admin from the company view.
5. Build a report with AI
   As the Admin, pick the vessel, type a job like “Enclosed space entry”, run AI generation, and walk the sourced hazards → consequences → mitigations. This is the wow moment.
6. Show the viewer experience
   Log in as a Standard user — view and download only, no editing. “Access is controlled by role.”

12Frequently asked questions

Does the AI make up regulations?

No. It only answers from the maritime documents in the Knowledge Base, and when the sources don’t cover something it says so instead of inventing — those non-answers are filtered out before they reach a report.

What if a vessel already exists?

The platform detects duplicates by name, IMO, or call sign and stops you from onboarding the same vessel twice — it points you to add a job to the existing vessel instead.

Can two people work on the same company?

Yes — a company can have many team members. Admins build their own reports; Standard users view all of them; the Company Master User oversees everything.

Who can download reports?

Everyone with access, including Standard users. Downloads are encrypted PDFs.

What are Q88, OVPQ, and VPQ?

Standard maritime questionnaires that describe a vessel. ADRSA-MARINE reads all three automatically to pre-fill a vessel’s details.

13Glossary