Aduara assigns every product its 10-digit HTSUS code, grounded in the live tariff schedule and CBP’s own binding rulings, reasoned through the General Rules of Interpretation — with the rationale, the rulings it cited, and a confidence score. A licensed customs broker reviews it before anything is filed.
Rather read first? Our guide to HTS classification and the GRIs.
Tell us what you import. We’ll reply by email and set up a short call to walk through your own lines.
It sets the duty rate, decides whether additional trade-program duties apply, and determines which partner government agencies get involved. Get it wrong and the cost shows up long after the goods have cleared.
A heading with a higher rate than the correct one means overpaying on every shipment — and unless someone catches it before the entry is final, it stays paid.
Importers must use reasonable care to classify their goods. Underpaid duty is owed back, and misclassification can bring penalties whose maximums rise with culpability — from negligence to gross negligence to fraud.
Section 301, Section 232 and IEEPA-era tariff actions are all scoped by HTS code. The wrong code can put you inside — or wrongly outside — an additional duty, and CBP audits look exactly there.
Every step below is documented in full in how the AI classifies goods.
A first pass restructures your free-text description into the attributes that decide classification — material, construction, function, user, presentation — and names what the description doesn’t say, so ambiguity is weighed honestly instead of guessed away. A product photo can be included when you have one.
For each line, Aduara retrieves the most plausible candidate codes from the live Harmonized Tariff Schedule — with a synonym map so “lamp” reaches “luminaire” — alongside the most relevant CBP CROSS rulings and worked examples of how CBP classified similar products.
Where headings genuinely compete, the classifier names them and decides in General Rules of Interpretation order — terms of the headings and notes, then most specific, then essential character — and cites the deciding rule and the controlling Section or Chapter Note in its rationale.
The code is checked against the live HTSUS — an unknown code is rejected. Today every AI classification routes to a licensed broker for review, and a low-confidence line blocks the whole entry until a broker has looked at it. Broker corrections become precedent for your future lines.
Drawback substitution matches imports to exports at the 8-digit HTS level, so the claim is only as good as the classification underneath it.
Explore duty drawback recovery →The HTSUS structure, the six General Rules of Interpretation with a worked example, CROSS rulings, and the mistakes that cost importers most.
Read the HTS classification guide →HTS classification is assigning an imported product its 10-digit code in the Harmonized Tariff Schedule of the United States (HTSUS). The first six digits follow the international Harmonized System; digits seven and eight set the U.S. duty rate; digits nine and ten are statistical suffixes. The code determines the duty rate, whether trade-program duties such as Section 301 or 232 apply, and which partner government agencies are involved.
It goes further than a lookup. A lookup returns codes whose descriptions match your keywords; Aduara reads the product description, retrieves the plausible candidate codes from the live HTSUS and the most relevant CBP rulings, decides between competing headings using the General Rules of Interpretation, and returns one 10-digit code with a written rationale, a confidence score, alternatives, and the rulings it relied on.
No. The AI proposes; a licensed customs broker approves and signs. Today every AI classification routes to broker review regardless of confidence, and a line below the review-block floor stops the whole entry until a broker has looked at it.
No. Every code the classifier returns is validated against the live HTSUS table, and a code that does not exist in the schedule is rejected and retried. If nothing fits, it says so with a confidence of zero rather than forcing a code.
For each line it retrieves the most relevant rulings from CBP’s CROSS database, plus worked examples of how CBP classified similar products, and cites the rulings it relied on in the rationale. A broker can check the basis instead of trusting a number.
The correction is recorded with the original code, the corrected code, and the reason, and it becomes precedent the classifier defers to on similar lines for your organization. Corrections are segregated to your organization by default and never pooled across customers.
Underpaid duty is owed back, and a misclassification can lead to penalties under 19 U.S.C. § 1592, whose maximums scale with culpability from negligence to fraud. Overclassification means overpaid duty you rarely get back. And because Section 301, Section 232 and other trade-program duties are scoped by HTS code, one wrong heading can add or remove an entire additional duty.
Against held-out CBP rulings: the ruling that would confirm the answer is excluded from retrieval, and each prediction is scored at the heading, subheading, and exact 10-digit level, with and without retrieval. The full methodology and current numbers are published in how the AI classifies goods.
Bring a handful of real lines — the tricky ones especially. We’ll walk through the code, the rationale, and the rulings behind it.
General information, not legal or customs advice. Classification depends on the specific facts of each product; a licensed customs broker reviews and signs every classification Aduara files.