Many of Smoker’s clients are Amish and Old Order Mennonite. Their financial records arrive as thick stacks of handwritten notepad pages — farm books, asset lists, income summaries — with no totals, no structure, and often barely legible penmanship.
Every number on every page had to be typed into a spreadsheet by hand. Interns spent six to seven hours per return on data entry alone. The firm went digital in 2014, but the workflow barely changed — staff scanned documents into Adobe and still stamped totals onto pages manually. The monotony burned people out and pushed returns past the April 15 deadline.
Lido pulls numbers and descriptions from Smoker’s handwritten client documents with near-perfect accuracy — even the worst penmanship. No more guessing whether a handwritten seven is actually a zero. Staff get clean, structured data they can work with right away.
Smoker also uses Lido’s AI to classify headers across 60+ payroll document formats and sort line items into the right tax categories — federal, state, local — automatically. The firm is building internal tools around Lido’s API: payroll classification workflows, partner notes digitization, and more. Partners write handwritten meeting notes, run the pages through Lido, and get clean digital summaries back.
Within the first two weeks of tax season, interns went from six to seven hours per return to two, at least tripling their speed. The firm filed significantly fewer extensions than the prior year. After a major accuracy update arrived mid-season, Andrew estimates 99.9% extraction accuracy, even on the worst handwriting.
The year before, Smoker relied on four outside vendors for tax prep review, HR, tech infrastructure, and document processing. Next year, Lido will be the only one. But for Andrew, the biggest change wasn’t speed; it was mindset. Lido was the firm’s first real exposure to what AI tools could do, and the team is now building their own internal tools on top of Lido’s API.
