By Luis Lopez, AI transportation consultant, CEO of Go Hub.io Holdings Corp and subsidiaries, and host of the Freight Guru Podcast
Most import pricing mistakes are not made at the customs window. They are made weeks earlier, when someone builds a quote using the supplier’s unit price, adds an ocean rate from an old email, and calls it a cost. By the time the duty, the fees, the drayage and the warehouse bill arrive, the margin on the product is already spoken for.
Landed cost estimation is the discipline of pricing an import with everything included before the goods ship. AI tools are getting useful at the tedious parts of that work: gathering quotes, organizing tariff information and running scenarios. They are not a substitute for a licensed customs broker or for the importer’s own legal responsibilities.
What landed cost includes
Landed cost is the total cost of getting a product to the point where it can be sold or used. For a U.S. import, the components generally include:
- Product cost: what you pay the supplier.
- International freight: ocean or air charges from origin to the U.S. port or airport.
- Cargo insurance: coverage for loss or damage in transit.
- Customs duties: based on the tariff classification and the customs value.
- Customs broker fees: entry preparation, filing and related services.
- U.S. government fees: such as the Merchandise Processing Fee (MPF) and, for ocean shipments, the Harbor Maintenance Fee (HMF).
- Drayage, warehousing and inland freight: moving the goods from the port to a warehouse and on to the final destination.
- Possible extras: customs exam fees, demurrage and detention, and additional tariff measures such as Section 301 or Section 232 duties where they apply.
Two points trip people up. First, for U.S. imports duty is generally assessed on the customs value, which is usually the transaction value, and that value generally excludes international freight and insurance. Confirm how it applies to your goods with a licensed customs broker. Second, the Incoterms rule in your purchase contract (ICC Incoterms 2020) decides which of these costs you pay and which the seller pays. If you are not sure what your terms mean in practice, start with Incoterms 2020 explained for importers.
Where AI helps
Pulling quotes into one place
Freight quotes, broker fee schedules, warehouse rate sheets and drayage rates usually live in emails, PDFs and spreadsheets. Document extraction and language models can read those sources and populate a structured cost template: origin charges, ocean or air rate, destination charges, chassis and accessorials, storage. It can also flag when a quote is missing a line you normally see. For a refresher on the charges that tend to hide in ocean quotes, see the top ocean freight charges importers should know.
Suggesting HTS candidates for broker review
Duty depends on the Harmonized Tariff Schedule of the United States (HTSUS) classification. AI can read a product description, specifications and photos and suggest candidate HTS codes with its reasoning. That gives your broker a head start and helps you estimate a duty range early in a sourcing decision. It does not give you a classification. Classification is the importer’s responsibility under the reasonable care standard, and a suggested code should always be reviewed by someone qualified. The deeper mechanics are covered in AI for customs classification and entry prep.
Flagging tariff changes to check
Additional tariff measures change often. A good landed cost tool keeps a record of which HTS codes and countries of origin your active products use, and raises a flag when something relevant may have changed so a person can check it against the current HTSUS and CBP guidance. The tool points a human at the question; it does not decide the answer.
Modeling port, mode and routing scenarios
Landed cost is not one number; it is a set of options. Ocean versus air, one port of entry versus another, transload near the port versus moving the container to an inland warehouse. AI-assisted models can run these scenarios side by side using your actual quotes, transit times and storage assumptions, and show how each choice changes cost and timing. This is where demurrage and detention risk belongs too. A cheap routing that regularly leaves containers sitting at the terminal is not cheap. If you need the background, read demurrage vs detention and how to avoid both.
Applying Incoterms consistently
Once a template knows the Incoterms rule for each purchase order, it can include or exclude cost lines automatically, so freight is neither double-counted nor left out.
The risks
- Classification errors. A confident but wrong HTS suggestion can understate duty for every unit you price. The importer, not the software, carries the responsibility.
- Stale tariff data. A model trained or loaded months ago may not reflect current duty rates or additional measures. Always know the date of the data the tool is using.
- Bad inputs. An expired freight quote or a wrong unit count produces a precise-looking number that is simply wrong. Precision is not accuracy.
- Hidden assumptions. If the tool assumes free time at the terminal, a particular warehouse rate or a specific port, those assumptions should be visible on the estimate, not buried in the model.
- Customs value confusion. Tools that compute duty on the total of product plus freight, without understanding how customs value is generally determined, will overstate duty. Have your broker confirm the basis.
A workflow that keeps a licensed broker in the loop
As an AI transportation consultant, the first thing I look at in a landed cost process is where the human checkpoints sit. A practical sequence looks like this:
- Collect product data. Description, materials, function, country of origin, supplier price and Incoterms rule.
- Let AI draft the estimate. Candidate HTS codes, extracted freight and fee quotes, scenario options, and a list of open questions.
- Broker review. A licensed customs broker reviews the classification candidates, duty basis, applicable additional measures and expected fees, and corrects the draft.
- Lock and version the estimate. Save the approved estimate with its data date, assumptions and the broker’s notes, so anyone can see what it was based on.
- Re-check before shipment. If tariff measures, freight rates or routing have changed between quote and booking, rerun the estimate and flag the differences.
- Reconcile after arrival. Compare the estimate to actual entry documents, freight invoices and warehouse bills. Feed the differences back into the template.
What to measure
- Estimate-to-actual variance by cost line, so you can see whether duty, freight or destination charges are the usual source of surprises.
- Classification change rate: how often the broker changes the AI’s suggested code. A high rate means the tool or the product data needs work.
- Missed cost lines: charges that appeared on actual invoices but were absent from the estimate.
- Data freshness: the age of tariff and rate data at the time each estimate was approved.
- Turnaround time from product request to approved estimate.
The bottom line
AI makes landed cost estimation faster and more complete by gathering quotes, suggesting classifications, flagging tariff questions and comparing routing options. The judgment calls still belong to the importer and a licensed customs broker. Use the tool to prepare the question and organize the evidence, keep the assumptions visible, and reconcile every estimate against what you actually paid.
For more conversations on freight and AI, subscribe to the Freight Guru Podcast.
About the author: Luis Lopez is a Miami-based AI transportation consultant and logistics entrepreneur, the CEO of Go Hub.io Holdings Corp and subsidiaries, and host of the Freight Guru Podcast.
