Every accounting firm I talk to is asking the same question: 'Should we be doing something with AI?' The answer is yes — but probably not what the vendors are selling you.
Let me separate what's real from what's hype, based on actually building and shipping AI features at Workday while also understanding how accounting firms operate.
What works today: document classification and extraction (invoices, receipts, bank statements), reconciliation assistance (matching transactions across systems), report generation (automated flux analysis, variance explanations), and knowledge-base Q&A (RAG over internal procedures and client files).
What's still experimental: fully autonomous bookkeeping, AI that replaces professional judgment in audit, and 'set it and forget it' automation. AI makes mistakes — confidently. You need human review on anything material.
The best place to start is a focused assessment of your top 3-5 time-consuming workflows, followed by a proof-of-concept on the highest-ROI one. Don't try to 'implement AI' across the whole firm. Ship one thing that saves real time, learn from it, then expand.
I do these assessments and builds for accounting firms and finance teams. The ones who succeed start small, measure results, and iterate. The ones who fail buy a platform license and hope the AI magically understands their clients.
This post is for general educational purposes only and does not constitute tax, legal, accounting, or investment advice. Consult a qualified professional for your specific situation. Content reflects the author's understanding as of the publication date; rules, rates, and interpretations may change.