If a client office, a board, or a professional indemnity renewal questionnaire has asked how your business uses AI, you've already met the moment this article is about. And if the question hasn't reached you yet, it's on its way, because AI reaches most maritime organisations before any formal adoption decision gets made. It arrives inside software updates, and it arrives when someone on the team finds a tool that saves an afternoon. That order of events is structural. The tools tend to turn up before the paperwork does, in businesses of every size.
A written AI position is the answer you can hand over when the question comes: what's in use, what it may touch, who reviews its output before it reaches a client or the sea, and which rules reach you. I want to be concrete about what that document does once it exists, so here are four engagements from my Log Book, with the numbers they produced.
The ferry operator with a sale in progress
A harbour cruise and passenger ferry operator: 22 vessels, around 500 staff, and 38 active systems across nine departments, the estate an operation accumulates when it grows over years and each need is met with the right tool for its time. The parent group, an international listed transport and tourism company, had the operation under active consideration for sale. Leadership could see AI ought to help somewhere, and they wanted evidence of where, an order to do it in, and a view on timing given the ownership transition in progress.
The engagement ran 18 to 20 hours of structured interviews with 13 stakeholders across all nine departments, alongside on-site observation and a full review of the technology stack. One finding gives you the flavour, and it's the pattern that comes with systems introduced at different times for different jobs: reconciliation moving between them by hand, in one role adding up to close to half a working week. Leadership holds that number now because they commissioned the review that found it.
What they hold now is a Blueprint with 11 sequenced AI and automation opportunities, each mapped to the systems it touches and the staff time it can recover, plus a custom AI tool that lets leadership interrogate the findings by department, opportunity, or system. Three questions governed the sequencing: which systems does this touch, which team owns it, and can the technology environment support it before the broader platform decisions are finalised. Leadership left knowing what to act on now, what to defer until after the transition, and what to present to an incoming owner as a documented plan. In a sale process, that last part carries a long way: the position itself became something the business can put in front of a buyer.
The shipbuilder with a tender threshold to meet
A commercial shipbuilder with more than seventy years of government and defence contracts, ISO9001 certified, with a small and highly skilled workforce. The methods behind those seventy years had held for decades: proposals assembled from the firm's own templates, technical manuals built by hand, job costing guided by experienced estimates, and eight operational systems each holding its own part of the record. Then one requirement moved: the Australian Government's Essential Eight cybersecurity framework became a threshold for defence tender eligibility, and the business asked for a formal assessment of its position against it.
The engagement covered factory walk-throughs across design, build, quality assurance, and handover, five stakeholder interviews, a review of all eight systems, and a formal Essential Eight baseline assessment. The deliverables: six prioritised opportunities, a 24-month roadmap in three phases, a 90-day action plan with defined success measures, the Essential Eight baseline with a remediation path, and a governance framework covering AI policy, role evolution, and change management. Every recommendation was tested against one question: could the business execute this without disrupting the delivery obligations that kept it viable? For a business tendering into defence supply chains, a documented position reaches all the way to which contracts it can bid for.
The charter agency whose best opportunity was already inside
A global superyacht charter agency: a small team, central agency and retail representation across international markets, and over 5,000 CRM contacts built up through years of charters and client relationships. The CRM held that database soundly, and no structured system was yet working on it, which is the usual state of a database built through years of relationship-led trading.
The assessment started with what the existing stack could already do before recommending anything new, because in a small-team operation the time cost outweighs the tool cost. Seven opportunity areas came out of it, sequenced into a three-phase roadmap: foundations first, then automation, then AI enablement. And the highest-return near-term opportunity was already sitting inside the business: return client re-engagement, with a 10 to 20 percent reactivation potential across the dormant contacts, requiring no new tools at all.
The team running several organisations at once
A compact team running several maritime organisations at once, with shared infrastructure but distinct identities and client bases, carrying the full operational load across all of them. The audit found seven operational themes repeating across the portfolio, the themes shared infrastructure produces when a compact team carries several identities at once: inbound communication, client capture, stakeholder reporting, and the structure of the CRM data.
The roadmap mapped ten strategic goals across two streams, workflow automation and AI-powered support, and the foundational recommendations are already implemented: online payments established, the CRM integrated with payment and onboarding workflows, and the manual follow-up on those flows gone. Foundations came before automation, and automation before AI enablement, because automation multiplies whatever it stands on, so the foundations get set first.
What the four have in common
Each of those organisations ended up holding a document someone else could act on. An incoming owner, in the ferry operator's case. A tender assessor, in the shipbuilder's. Their own team next quarter, for the charter agency and the portfolio. The governance in each engagement was the standard the result was delivered to, the thing that makes the plan defensible when a board, a buyer, or an authority reads it, and the benefit shows up in what it let each business do next: present the plan, enter the tender, run the reactivation, switch on the payments. And when the plan's next line item is a build, what a finished AI build looks like shows how that stage of the work runs.
MIT's research on the GenAI Divide found 95 percent of enterprise AI pilots deliver no measurable return, and the reasons are rarely technical: misaligned priorities, poor sequencing, no executive ownership. A written position is how you end up in the other five percent, because it settles what happens first and who owns it before any money moves.
On the insurance side, if you're renewing professional indemnity or cyber cover, you may notice the proposal forms now ask where AI sits in your business. How an insurer weighs your answer stays the insurer's decision. Having an answer ready in writing is the part you can decide today.
Could you answer these three in writing, today?
- Which AI tools are in use across your business right now, including the ones that arrived inside software updates?
- Where does a person check the machine's output before it reaches a client or the sea?
- If a client, a board, or a renewal questionnaire asked for your AI position this week, what document would you send?
If all three came easily, your position may be closer to documented than you think, and writing it down is a smaller job than it sounds. If one of them stalled, that's the common starting point, and it reflects how AI arrives rather than how anyone runs their business.
The plain place to start is the AI Baseline: twenty short questions, about five minutes, and a written read of where AI already sits in your organisation, what it exposes, and which rules reach you. From there, Governance Essentials is a guided process that produces your AI Use Policy, the document a client, an insurer, or a lawyer will eventually ask for, written to the regimes that apply to you rather than from a blank template. And for organisations that want the full picture, the Blueprint runs leadership interviews, risk and oversight mapping, a policy your board can sign, and a sequenced roadmap of what to do first.
Get your baseline at southernsky.ai/baseline. If you'd rather talk it through first, request the Engagement Guide.
