Use a narrow maritime AI prototype to test real US desk demand

A prototype should test a customer problem

A maritime AI prototype is most useful when it tests whether a specific US desk problem is worth solving. A broad demonstration can impress users without revealing whether they would change their workflow. A narrow prototype can show whether the proposed assistance helps with a recurring task, fits the user's evidence, and earns another use.

YC's official How to Talk to Users lesson covers interviewing current and potential users and interpreting feedback. NSF's I-Corps Teams description frames entrepreneurial learning around evaluating commercial opportunity. These are educational anchors, not endorsements of a maritime company. The experiment design below is original analysis for US shipbroking and cargo workflows.

Zeaclub's founder identifies YC as an inspiration. That can be expressed credibly through a focused learning process: choose a real problem, observe users working through it, and revise the proposed solution based on evidence. It should not be presented as YC affiliation or as proof that the product already solves the problem.

Choose a task with an observable outcome

A prototype needs a task users can recognize and evaluate. Prepare a cargo enquiry summary is more testable than transform shipbroking. Compare quotation scope is more concrete than improve decision intelligence. The task should have a visible result and a clear point at which the user can say whether it helped.

Select a recurring problem rather than an unusual case that generated a memorable complaint. Frequency matters because a useful one-off tool may not justify a continuing product workflow. At the same time, do not ignore rare consequential cases entirely. They can be included later to test the limits of the proposed role.

Keep the prototype's boundaries explicit. It may work only with supplied documents, prepare internal output, and require human review. Those limits are acceptable if they match the question being tested. The prototype should not imply live integrations, automated sending, or verified vessel availability unless those capabilities actually exist and are part of the trial.

A fictional enquiry-summary experiment

Imagine a hypothetical US brokerage that often receives incomplete machinery enquiries. The proposed prototype extracts the supplied particulars and prepares a concise list of missing information. The experiment asks whether this helps brokers reach a useful clarification request faster than their existing process.

The team selects redacted examples from ordinary work, including one with complete information, one with missing dimensions, and one with conflicting updates. Users perform the task while observers note where they inspect sources, edit the output, or abandon the prototype. The exercise does not require the prototype to select a vessel or negotiate a fixture.

A user says the summary looks excellent but then rewrites every clarification question before using it. That is different evidence from a user who corrects one field and asks to use the tool on another enquiry. Compliments describe an impression. Observed behavior shows whether the output fits the work.

Suppose users consistently value the missing-information list but ignore the generated narrative. The next iteration should investigate that pattern rather than add more prose. The useful product may be smaller than the original concept. A narrow experiment can reveal where the value actually sits before the team invests in a broader interface.

Ask about recent work, then observe use

A customer conversation should establish the existing process and its friction. Ask the user to describe a recent enquiry, what they did, where they lost time, and what happened next. Avoid relying only on questions about whether they would like a hypothetical feature. People can find an idea appealing without having a strong reason to adopt it.

When showing the prototype, give the user a real task rather than a guided tour that tells them which buttons to admire. Observe whether the output supports their next action. Ask why they change or reject a suggestion, and distinguish missing functionality from an incorrect understanding of the workflow.

Do not treat one user's preference as a universal requirement. A broker, an operator, and a shipper may need different outputs from similar source material. Record the role and task context alongside the feedback. The prototype should help clarify which customer and use case the product serves first.

Define evidence that would change the plan

Before the trial, state what the team expects to learn and what result would cause a revision. For example, the prototype may need to reduce reconstruction work without increasing material errors. If users spend more time verifying the summary than preparing it manually, the design or task selection needs attention.

A trial should also reveal whether the problem is primarily technological or procedural. If users cannot agree on the meaning of cargo ready, the product may need a clearer workflow definition before more automation. If source documents are consistently incomplete, a better intake process may matter more than a stronger model.

Keep the evidence proportional. A few controlled sessions can identify usability problems and promising behavior. They do not establish a market-wide productivity claim or a guaranteed commercial return. The next step should follow the uncertainty remaining, whether that is another user group, a different input format, or a longer period of ordinary use.

Separate opportunity narrative from product proof

Zeaclub's founder describes hundreds of millions of dollars in commercial opportunity pipeline, meaning potential transaction value rather than revenue, signed fixtures, or funding. That narrative can explain why the business is interested in maritime workflows. It does not substitute for evidence that a particular prototype helps a broker complete a task.

The distinction is productive rather than limiting. A large opportunity can motivate exploration, while a small experiment can show where to begin. The strongest customer discussion connects the broader ambition to one problem the user can demonstrate and one result the product team can evaluate honestly.

For a US desk considering a conversation with Zeaclub, bring a redacted example of recurring work and a clear description of the current process. Ask to evaluate a narrow proposed workflow against that example. The useful outcome may be a promising prototype, a revised problem definition, or a decision that another approach is more appropriate.

A prototype earns further development when observed use supports the next investment of effort. Start with the customer problem, preserve the limits of the experiment, and let actual behavior guide the next version. That approach turns YC-inspired learning into a concrete maritime product practice without borrowing an affiliation or inventing evidence of success.

Sources

Related articles