109 articles
In four weeks, AI disclosure went from ethics debate to infrastructure: Substack scans it, LinkedIn flags it, the EU legislates it, and Anthropic watermarks everything Claude writes. Your firm almost certainly faces no legal risk, but your work product now carries flags you can't remove. The only question left is whether the story that arrives with the flag is one you wrote.
Xero's AI has coded 100 million transactions, and practitioners trust exactly one of its four methods: the one built from written rules. Every coding engine on the market infers your client's intent from history. Not one of them will read it, even though you could write it down today.
Intuit is bundling its own human bookkeepers into QBO Advanced, one toggle away in your client's settings. The same week, a CPA-graded benchmark scored AI models higher for disclosing uncertainty than for being right, and the EU wrote human editorial responsibility into law. Disclosure just became the product, and the record of who stood behind the number is what's left to charge for.
I have had the conversion moment twice. In 2024 I saw exactly what AI was going to do to this profession, spent six months learning it, and trained my team. Busy season came and everything went back to how it was. In early 2026 I did it again, with the same result. Here is what those two failures taught me about the only test of transformation that cannot be gamed.
Whether your firm should build or buy was never about the app. It's about whether you can supply the harness and the janitor yourselves. Part 3 of Buy vs. build closes the series with the build-readiness test, the packaging premium, and one question per layer to ask before any renewal.
Basis says it takes responsibility for the first pass of your clients' tax returns. OpenAI, four days earlier, refused to claim authorship of work its own model did and took responsibility only for its correctness. Both sentences are about who is answerable, and only one of them means it.
In the late 1990s I helped build a global audit platform for a Big Six firm. The capability let us do some old things faster, and start doing things we should always have been doing and never could. Quality went up. The clock never moved. Leadership was watching the clock, saw nothing, and cut the budget. If you are measuring AI in hours saved, you are about to read the same zero.
Per-client and per-user app fees levy a decision tax accountants have paid since QuickBooks Desktop gave way to subscriptions. Part 2 of Buy vs. build reads the pricing wrapper the way an accountant should, as a risk contract: who carries the token volatility, who holds the repricing option, and what to watch as usage-based pricing arrives.
Karbon asked clients what should happen to your fee when AI makes the work faster, and 37% said it should come down, which is close to what the IRS told tax practitioners in June. Deliver something they have never had instead and 54% will pay more. Three of the four walls of the pricing vise moved in a single week.
Xero has had a request open since 2017 to let you export bank rules, with 540 votes and still nothing shipped. QuickBooks Online has done it for years. As your firm starts encoding real judgment into AI agents and skills inside vendor platforms, that gap stops being a nuisance and becomes the question that decides whether your build compounds for your firm or evaporates.
June's conferences said buy and build in the same breath, and agreed most accounting apps are LLM wrappers. But the intelligence was never the whole product: every app subscription bundles five layers, and the wrapper critique prices only one. Part 1 of Buy vs. build unbundles what an app actually sells you.
Anthropic shipped two things this week that answer the same question in opposite ways: a feature that turns one screen recording into an editable procedure file, and a model that checks its own work brilliantly and remembers none of it. Meanwhile OpenAI launched a program teaching small-business owners to do their own accounting, and a top-15 firm committed its whole client accounting practice to a firm-trained AI model. The question running under all of it is which of these leaves anything behind.
