TXF Intelligence

Taxfyle built its business as a marketplace matching returns to preparers, and TXF Intelligence is the software layer it now sells to firms directly. The loop is short: bulk upload a client's documents, the system categorises and organises them, extracts the figures from W-2s, 1099s, receipts and whatever else arrived, presents the result in a review screen the preparer checks, and then generates a return that transfers into Drake or Intuit ProConnect. The claim on the page is that this removes more than 40 percent of manual data entry.

Pricing is stated openly, which is rare enough here to matter, and it is structured as annual credit blocks rather than a subscription. Starter is $3,000 a year for 100 credits at $30 a return, Growth is $10,000 for 400 credits at $25, and Premium is $20,000 for 1,000 credits at $20. Seven days of full access plus five credits comes free. Do the arithmetic before assuming that is cheap: at 200 returns a firm is buying Growth and using half of it, so the effective cost is $50 a return, and the advertised per-return figures only hold if the credits are consumed. At 1,500 returns the Premium block covers two thirds of the volume. Compared with StanfordTax at $18 a return with credits that roll over, this is the more expensive unit, and it is committed up front.

Two things should slow a buyer down. Only Drake and ProConnect are supported, which rules out any firm on UltraTax, Lacerte, Axcess or GoSystem. The social proof is also unverifiable: three testimonials that name no firm a reader could look up, alongside accuracy numbers of 99.8 percent set against 85 to 90 percent for seasonal temps, with no method published behind either figure. The page states SOC 2 Type II certification, US-based processing and role-based access, but publishes no security page to substantiate it. Review is described as the preparer checking a clean version of the extracted data, so the human queue is the firm's own staff rather than a vendor verification team, which is honest but also means the labour saving is capped at what a reviewer can confirm quickly.

Against Filed, Byron and Magnetic, which sell AI preparation as a service rather than a licence, TXF asks the firm to keep the review work and pay per credit. Against Soraban the overlap is document intake, where Soraban is broader on the client-chasing side. Against StanfordTax the comparison is direct on price and unfavourable, though StanfordTax stops at the binder while this writes into the tax program.

One-to-watch is the right place. Transparent pricing, a plausible workflow and a parent company with real volume behind it are all points in its favour. What is missing is the evidence that would move it up: named firms using it through a full season, a published security page, a second tier of tax software support, and references a firm can actually contact.