Tarifflo (Logan, UT), an AI startup from Utah State’s Sandbox cohort, automates tariff classification using agentic AI. The company won the $20,000 Grand Prize at the Tim Draper Utah Entrepreneur Challenge.
Logan, Utah - March 21, 2025
Tarifflo, an AI tech startup from Utah State’s current Sandbox cohort, aims to automate tariff classification of internationally imported and exported products, which can be a very expensive, time consuming process. By leveraging new agentic AI model’s, Tarifflo is able to use software to accurately automate tariff code classification.
[Editor's Note: Tarifflo won the $20,000 Grand Prize at the 2025 Tim Draper Utah Entrepreneur Challenge, announced on March 22, 2025 at a final awards and showcase event at the University of Utah. The statewide business-model competition is managed by the Lassonde Entrepreneur Institute, an interdisciplinary division of the David Eccles School of Business at the University of Utah, and sponsored by Tim Draper, a renowned venture capitalist. The final event featured the best collegiate entrepreneurs in Utah competing for more than $75,000 in cash and prizes.]
Tarifflo automates import documentation, outperforming traditional and competing software solutions. Their proprietary technology uses product information and generates detailed documents, including comprehensive references to customs rulings and regulations. They also offer tariff-rate management and possess a large trade-compliance database covering non-tariff measures, free-trade agreements, country-specific harmonized tariff schedule codes and indirect taxes.
Tarifflo Founder Tanner Helms further explained the problem that Tarifflo solves:
“To classify a product, it costs $100 on average at a customs brokerage, and it can take them a few weeks to get back to you with the data. It can be expensive and slow. Even going through a licensed custom broker, about 30% of classifications are still incorrect.”
He continued:
“Incorrect classifications, even just one incorrect code can lead to unexpected fees, shipping delays and steep fines.”
Tarifflo also already has funding despite being such a new company, as Helms stated:
“We closed our pre-seed round in January. We raised a bit over $500,000 on a $10 million valuation cap. That has already made a big difference, allowing us to scale up a lot quicker and hire developers.”
The market for international tariff classification is quite large, as Helms explained:
“Internationally, products require a 10 digit HTS code that's used to determine the tariff or duty rate. Right now, there's over 400,000 companies importing products each year, which equates to about $4 trillion worth of goods being imported for a year.”
Helms pointed out how steep the fines can be:
“Last year, Ford Motor Company was fined $365 million for using one single incorrect tariff code. They were importing cargo vans and they were using the HTS code that has a 2.5% tariff rate instead of a 25% tariff. They were actually underpaying in tariffs, and got hit with a huge fine for that.”
With constantly changing tariff laws, sometimes companies have to temporarily shut down to adjust to the new laws. Helms noted:
“On February 1st of this year, the De Minimis rule was revoked. The De Minimis rule allowed a lot of shipments under $800 to pass through customs without paying tariffs or having an HTS code. This now means an additional 1 billion imports per year from the US, Mexico, and Canada need to have full HTS classifications. This had such an impact, that it has temporarily been put back into place allowing some additional time for companies to complete their classifications and documentation.”
He added:
“When that originally went into place, USPS stopped accepting packages from Hong Kong and China temporarily because they had to figure out what they were going to do about this change."
Tarifflo believes they have the software solution to the many problems tariff classification presents.
Helms explained how the software works:
“What we’ve developed is an agent that thinks in the same way a human would when going through the classification process. In order to keep our database up-to-date, every single night our AI agents run, pull the new updates, update our database, and then use that data to make classifications.”
Tarifflo is easy to use from a user perspective, as Helms describes:
“The user can upload products by either a CSV file or by manually entering a product one by one with a name and description. Once they’ve done that, they’re done, and our system finds the HTS codes for them.”
He continued:
“Once the classification is completed, they have a dialog that pops up on the screen where they can view all the information needed for that product.”
Tarifflo also provides quick response times, especially compared to traditional means of tariff classification:
“Our system can auto scale up, giving us constant response times no matter how many products are uploaded. Whether it's 500 products or 10,000 products, it takes roughly 200 seconds for the entire process to complete,” Helms stated.
Pricing for Tarifflo was provided by Helms:
“It’s a SaaS, B2B tier pricing model. Our plans start from $2,500 per year up to $75,000 per year depending on the client size and item volume. We also offer free audits to see if your company has any compliance risk or if you’re underpaying or overpaying in tariffs.”
The idea for Tarifflo was started by Co-Founder Bryce Judy, who used to perform tariff classifications himself, as Helms explained:
“My co-founder Bryce worked at KPMG doing HTS classifications by hand. He was doing classifications for Dick's Sporting Goods, as well as some other large companies. While working there he found that the process is essentially the same order of operations for every single type of product, and it's easily repeatable.”
He continued:
“But, the issue was that they tried to automate this process using machine learning, and typical machine learning requires you to train some neural network on past data. But the problem is that when there's so many tariff updates happening so frequently, the update is changing so fast, making the neural network out of date and leads to classification errors. Only in the last year AI agents have become sophisticated enough to take in real-time data and be able to automate classifications.”
Helms and Judy met through the Sandbox program, and from there have been able to grow their company, as Helms said:
“Me and Bryce met through Sandbox. He initially contacted me and had a different idea at the time, but we then pivoted to going through with the Tarifflo idea. Since then, we’ve grown a lot and have gained a lot of support. We currently have 15 people working for Tarifflo, our investor network is about nine people, and we also have a lot of mentors providing tons of support.”
In the future, Helms stated Tarifflo will automate everything, including full document creation:
“The big thing happening soon is we're working with US Customs and Border Protection, to actually fully automate the documentation of entry summaries to the government and on a company's behalf.”
Tarifflo also plans to start working with World Trade Centers in different states, as Helms explained:
“We’re currently in talks with four World Trade Centers, meaning we’re in talks with four different states. If we can generate a partnership with these World Trade Centers, that’d be really big for us, because then we’d be a referral standpoint for anyone going through that trade center.”
Learn more about Tarifflo by clicking here.