US LNG Enquiry Automation: Keep Conditional Cargoes Out of the Firm Supply Column

Extracting a quantity must not strengthen the offer

An automated US LNG enquiry workflow should preserve the conditions attached to cargo availability. If a seller describes a quantity as subject to confirmation, the output should retain that status. A system that extracts the number correctly but drops the condition has changed the commercial meaning of the message.

NIST's generative-AI profile identifies confabulation as a risk involving erroneous content. Fluent output therefore needs evidence-based review before it is used in a consequential commercial decision. NIST generative-AI profile

For LNG enquiries, the problem is often more subtle than an invented vessel or port. It can be an unsupported upgrade from possible supply to firm supply, or from an annual entitlement to a specific cargo. The automation should make those distinctions visible so the commercial team can decide what further enquiry is justified.

Treat the qualification as part of the field

A useful cargo quantity field includes its unit, time basis, status, and relevant condition. The quantity is not complete without them. Annual energy volume, a proposed lifting quantity, and an optional increment answer different questions even when the same unit appears in each.

A public Port Arthur Phase II contract summary qualifies its stated annual quantity by operational and partial-year conditions. It demonstrates how a number can depend on surrounding text. DOE-hosted contract summary

That document does not establish present cargo availability. Its value for this design question is the relationship between the numerical provision and its qualification. An extraction workflow should capture both, preserve their source, and avoid converting the annual figure into an assumed sequence of liftings.

The same principle applies to loading dates. A target month, a proposed range, and a confirmed window should remain distinct. If the source only provides a month, the system should not invent a narrow window because its downstream form expects exact dates. It should identify the missing information and route the question to the appropriate person.

A fictional conditional-quantity test

Imagine a fictional US LNG enquiry stating that 2.5 million MMBtu is confirmed, with a further 500,000 MMBtu possible if the seller completes an internal allocation. A second message says the optional amount remains under review. The correct commercial summary contains a firm quantity and a separate conditional quantity, not one firm total of 3 million MMBtu.

If the buyer's customer needs 3 million MMBtu, the unresolved supply gap is 500,000 MMBtu. At an invented $11 per MMBtu replacement assumption, that gap represents $5.5 million of gross replacement purchase exposure before other costs. This is an illustration of why the condition matters, not a current price or a prediction that replacement will be required.

The business may still choose to discuss shipping for the larger quantity. Its enquiry should explain the option and the decision date, allowing counterparties to respond on the intended basis. The error would be to present the full quantity as settled and later discover that the carrier proposal or customer commitment depends on supply that was never confirmed.

If a later authorized message confirms the increment, the system can update the status while retaining the earlier source history. If the message merely asks whether confirmation is possible, it should remain a question. The workflow must distinguish a request, an answer, and an internal interpretation.

Preserve contradictory evidence rather than choosing silently

Commercial messages can conflict. A seller's earlier indication may state one quantity while a later attachment gives another. The automation should identify the conflict and show the relevant sources rather than select the larger number or the most recent timestamp without understanding the relationship.

A later message may discuss only part of the transaction. It might revise timing while leaving quantity unchanged, or refer to a different cargo entirely. Useful extraction therefore links the statement to the correct enquiry and its subject. It should not treat every message in a conversation as a complete replacement of every previous field.

Human review should focus on changes that alter the commercial position: quantity becoming firm, a condition disappearing, a destination changing, or a payment obligation being introduced. The reviewer needs the underlying passage and enough surrounding context to assess the proposed interpretation. A confidence score without that evidence is insufficient.

Test the workflow with meaning-changing cases

A practical evaluation set should include conditional quantities, withdrawn proposals, unanswered questions, different energy units, annual figures, and multiple cargoes in one thread. The expected result should specify the commercial meaning, not merely whether the software found the same digits as a human.

For example, a test can fail when the quantity is correct but the status is wrong. Another can fail when a condition is captured but attached to the wrong cargo. These are meaningful failures because they can change the shipping enquiry or customer promise. The tests should reflect actual decision consequences rather than reward a visually complete form.

The workflow should also allow an unresolved output. If the evidence cannot establish the firm quantity, returning an explicit unknown with a source question is better than filling the gap from common practice. Automation creates value when it reduces review effort while preserving the boundaries of what the source establishes.

Zeaclub's direction toward more intelligent maritime workflows can be expressed through this kind of practical design goal: structured enquiries that retain conditions and make important uncertainty easy to review. This is a proposed approach, not a claim that a specific automated capability is already deployed. Bring Zeaclub a representative LNG enquiry and the decisions your team needs to support so the discussion starts with commercial meaning rather than a promise of automatic certainty.

Sources

Related articles