Fran's Chocolates Tests Avalara's Agentic AI for Tax Compliance
15m
Small companies are finding that agentic AI for tax compliance can do more for them than it does for the big guys. At Avalara CRUSH 2026, Alan Shimel talks with Ken Harrison, senior accountant at Fran's Chocolates, about sales tax compliance after the Wayfair decision and his time as a beta tester for Avalara Aviator.
Fran's Chocolates: handmade at national scale
Fran's Chocolates is a family-owned chocolatier in Seattle that still makes, packs and ties every box by hand. This year it will produce more than five million pieces of chocolate. Online and mail order sales can reach 40 to 50% of revenue, so the company ships to customers across the country.
Why Wayfair changed tax compliance
The 2018 Wayfair Supreme Court decision let states collect sales tax on interstate sales. As a result, sellers suddenly had to track thresholds and rates in every state and even local jurisdictions. Harrison knew a two-person accounting team could not manage that alone, so Fran's turned to Avalara. Today it uses AvaTax, Returns, ECM and the new 1099 product to keep its tax compliance on track.
Beta testing agentic AI with Avalara Aviator
As an Avalara ambassador, Harrison volunteered to beta test Aviator, the company's new agentic AI product. Over three to four weeks he spotted a tax compliance anomaly in a cash report, asked Aviator where it came from and traced it to a single document. In the end, the problem was the data Fran's had sent, and his team fixed it at the show.
Agentic AI as a tax compliance force multiplier
At Fran's Chocolates, Harrison calls himself a mid-beginner with AI because finance data has to stay secure. Still, he sees agentic tools as an investment that helps a small team work smarter, catch issues and pull more insight from its reports.
Small businesses can get outsized value from agentic AI in tax compliance work. Explore more Avalara CRUSH 2026 coverage and the latest Techstrong TV interviews.
For more information please visit frans.com