THE PROBLEM
Classifying imports well is slow, expert work.
Five levels, one code
Chapter, heading, subheading, tariff item and statistical code decide the duty rate and which biosecurity, concession and dumping checks apply.
Notes, rules and splits
Brokers read chapter and section notes, apply the General Rules of Interpretation and settle the statistical split for every line.
Unchecked AI adds risk
A confident-looking but invented code can be accepted when nobody checks it.
THE AI TARIFF CLASSIFIER DIFFERENCE
From second-guessing to signing off with confidence.
Without it: pressure and doubt |
|---|
A blank page for every line, with five levels to work through. |
Second-guessing whether the code will hold up. |
Worry that an AI will hand you a confident code nobody checked. |
Routine lines take the time that borderline goods deserve. |
Reasoning that lives in your head and your notes. |
With it: control and confidence |
|---|
A checked first pass on every line, usually in under a minute. |
Confidence shown at every level, and any weak level named. |
A code MOSAIC's tariff data doesn't list is caught, and the result says so. |
Routine lines move fast, so your time goes to the borderline goods. |
One PDF with the code, alternates, rationale and sources for the job file. |
BUILT FOR BROKERS
Built for brokers, because the broker decides.
A faster first pass, with you firmly in charge of every code that goes to lodgement.
Speed on routine lines
A first-pass code usually in under a minute, with batches of lines handled together.
Checked, not invented
Codes are checked against MOSAIC's tariff data and the official wording.
Your expertise decides
Where the tool can't decide, it says so and asks you or holds the result for review.
A record for every job file
The code, alternates, rationale and sources in one downloadable PDF.
Examples on test goods. Results are indicative and not a formal Tariff Advice.
FAQ
Does it replace my customs broker?
No. It gives a qualified broker a checked first pass. The broker reviews the result, makes the final decision and lodges.
What if it can't decide?
It says so. The result comes back Provisional with the short level named, and MOSAIC asks you a question or holds the result for review.
How do I know what to trust?
Every result shows its confidence at each of the five levels. Fast mode settles a level at 80% and Expanded at 95%. Codes are checked against MOSAIC's tariff data and the official wording, and you verify before lodging.
How do I get the best result?
Treat it like briefing a colleague. Include what the goods are, what they're made of, key measurements with units, what they're used for, and a brand, retailer or link if you have one.
What happens to my data?
Descriptions are processed by MOSAIC's AI service. Classifications aren't saved to your MOSAIC records, so download the PDF report before you leave the page. Working copies are deleted 7 days after their last activity.
NEXT STEPS
See it on your own goods.
Book a demo and we will run the AI Tariff Classifier on your own goods, with your broker alongside.
1
See a sample result
A labelled sample result in MOSAIC.
2
Run your own goods
Classify a handful of your real products in Fast and Expanded, with your broker alongside.
3
Switch it on
Yojee turns the Tariff Classifier on for your organisation so your team can use it on live jobs.
Guidance only, and the broker decides. Results are not a formal Tariff Advice, so verify the code, duty and BICON against the current ABF Working Tariff before lodging.