Skip to content
upshift
Accounting Firms

AI handles the bookkeeping now. The question small businesses ask next is why they still need a CPA.

Clients still need a CPA for the value a firm adds beyond the bookkeeping. In a late-2025 Source Global Research survey of more than 50 senior tax buyers at large US companies, 66% expected to pay less for the same advice delivered with AI. Every one of them expected to pay more if an AI-enabled firm delivered more value.

What’s happening

AI-powered bookkeeping and tax preparation tools are improving fast. Small business owners can now categorize transactions, reconcile accounts, and generate financial statements using tools that cost a fraction of a monthly accounting retainer. The output isn’t perfect, but it’s good enough for a business owner who’s watching every dollar.

The threat is not that AI replaces CPAs entirely. It’s that AI handles enough of the routine work that clients start questioning the scope, and the fee, of the engagement. When a client’s AI tool categorizes 95% of transactions correctly, they wonder why they’re paying a firm to do 100%. The 5% gap doesn’t justify the full fee in the client’s mind.

This is the same dynamic that disrupted tax preparation for individuals. TurboTax didn’t eliminate CPAs, but it moved the threshold for ‘I need a professional’ dramatically upward. The same shift is happening for business accounting, and it’s happening faster because the AI tools are better. In Thomson Reuters’ 2026 survey, 74% of corporate tax departments want their outside firms using AI, but most don’t know whether their firms actually are. The demand is there; the gap is in telling clients what the firm offers.

There’s a second front: clients increasingly arrive with answers instead of questions. Practitioners trade stories about taxpayers showing up with AI-drafted positions. One described a client who had planned a chain of 1031 exchanges around bonus-depreciation math ChatGPT got wrong, then defended the plan against the professional correcting it. The client didn’t stop being wrong. He stopped assuming the professional must be right. Once that deference goes, fee resistance follows close behind, because a client who believes he can check your work believes he knows what it should cost.

Where this sits: The Three Shifts

Why the obvious responses don’t work

“Educate clients on risks of AI-only accounting”

Fear-based retention doesn’t last. Clients who stay because they’re afraid of AI errors will leave the moment they trust the tools, and the tools are getting more trustworthy every quarter.

“Bundle more services to increase switching costs”

Bundling commodities doesn’t create value. If each individual service is being commoditized by AI, packaging them together creates a bigger commodity at a higher price.

“Compete on relationships”

Relationships matter, but they don’t justify fees alone. A client who likes their accountant but can get 90% of the work done by AI will eventually make the rational choice.

“Win the argument when a client brings ChatGPT’s answer”

You can correct the model’s mistake and still lose ground. A client who cross-examines your advice against a chatbot has already reclassified you from trusted adviser to vendor under audit. What restores the standing is an engagement built around work the tool can’t do, where the client watches the judgment operate.

What’s working instead

In a late-2025 Source Global Research survey of more than 50 senior tax buyers at large US companies, 66% expected to pay less for the same tax advice delivered with AI. Every one of them expected to pay more if an AI-enabled firm generated more value as a result. Clients won’t pay more for AI doing what they already pay for. They will pay for what it lets you deliver that you couldn’t before.

The way out is to stop optimizing the old model and sell new services built on what AI can’t do. We find out what your clients will pay for, then build the AI that delivers it. Start with a pilot: in two weeks, a working first version tried on your own client work, then a decision on production. Or start with a plan for the whole firm.