Why aren't accounting firms using AI every day if the tools are so good?

The obstacle is rarely the tool — it's the structure around the tool. Accountants who tried AI once or twice usually did it without a specific workflow in mind, without any team training, and without anyone explaining why the firm was adopting it. Without that structure, AI use is optional, inconsistent, and easy to abandon when work gets busy. The firms that move from occasional to daily use have done three things: they picked one specific workflow to start (not a general experiment), they ran actual training sessions, and they set clear expectations about when and how AI should be used on client work.

What is the easiest first AI workflow for a small accounting firm?

Meeting summaries are the lowest-friction starting point for most accounting firm teams. Use Fathom or Otter.ai (both have free tiers) to record and transcribe client calls, then generate a structured summary of action items and follow-ups. This adds zero time to existing workflows, produces immediate visible value, and requires no review of confidential tax data or financial statements — which removes the compliance anxiety that blocks adoption in other areas. Once the team is comfortable with AI-assisted meeting documentation, move to the next workflow: client email drafting from notes.

How do you get staff to actually use AI tools instead of going back to their old workflow?

The research is clear: 17% of workers globally say their employers invest in the skills they need for AI. When employers don't invest, workers don't change habits. The minimum effective training is a 90-minute hands-on session per team, not a PDF or a shared link to documentation. The session should cover: what the tool does, what it doesn't do, when to use it for client work and when not to, and what review step is required before output goes to a client. After that session, designate one team member as the 'AI lead' who answers questions and tracks what's working. Without that structure, adoption reverts to the mean within 60 days.

What does 'human oversight' mean in practice for AI-generated accounting work?

It means a licensed professional reviews all AI output before it reaches a client or is filed with any regulatory authority. Specifically: any AI-generated draft tax return, financial summary, or client advisory memo must pass through a review step by a CPA before delivery. This is not optional — it's your professional and ethical obligation, and increasingly it's a regulatory one too. In practice, set a firm policy that AI output is always a first draft, never a final deliverable. Build the review step into your workflow explicitly, not as an afterthought. Some firms add a brief note to client-facing AI-assisted documents: 'Prepared using AI tools with professional review by [CPA name].' This is good practice for transparency and increasingly expected.

Is AI adoption different for smaller accounting firms vs. large ones?

Yes — significantly. Large firms have IT departments, innovation teams, and budgets for enterprise AI platforms. Small and mid-sized firms (5-50 employees) are working with more constrained resources and more personal client relationships. The advantage for small firms: you can move faster. There's no procurement committee, no multi-year software contract, and no enterprise rollout to manage. A 10-person firm can test, adopt, and embed a new AI workflow in two weeks if the owner decides to. The disadvantage: you also have less margin for error, less dedicated time for training, and no one whose job it is to stay current on AI tools. That's why starting with one specific workflow — not 'AI strategy' as an abstract project — is the right entry point for a small firm.

AI Daily Use at Accounting Firms: The 4-Step Adoption Framework

Published: March 17, 2026 | By: The Crossing Report | 6 min read

Most accounting firm owners have tried AI.

Most of them tried it a few times. Asked ChatGPT something, got a reasonable answer, maybe used it for a draft email. Then got busy, went back to the usual workflow, and haven't opened the tab since.

That's not a technology problem. That's an adoption problem — and it's so common there's now data on exactly how widespread it is.

Only 19% of accounting professionals use AI tools on a daily basis. That's one in five. The rest are either stuck in occasional-use mode or haven't started at all — 17% report they have never used AI at work.

This comes from ADP research published by CPA Practice Advisor in March 2026, with commentary from Andrea Wynter, VP at ADP Canada. The gap she's documenting is not about access to tools. The tools are cheap and widely available. The gap is structural: most firms have no framework for moving their teams from "tried it once" to "uses it every day."

Wynter's research identifies four specific steps that close that gap. They are not complicated. They are also not obvious — which is why most firms skip them and wonder why adoption stalls.

Summary

Only 19% of accounting professionals use AI daily — despite widespread access to affordable tools. ADP research identifies four structural reasons adoption stalls and four corresponding steps to fix it. For accounting firm owners, the path from occasional experimenter to daily user is not about buying a better tool. It's about building the framework around the tool you already have.

The Gap That Isn't Closing on Its Own

There's a persistent assumption that AI adoption is just a matter of time — that once the tools are good enough and affordable enough, everyone will use them naturally.

That assumption is wrong.

The accounting profession has had access to capable, affordable AI tools for two years. Adoption hasn't followed a smooth curve. Instead, there's a bimodal distribution: a smaller group of early movers who embedded AI into their daily workflows, and a much larger group still in pilot mode. The 1-in-5 daily use rate reflects that divide.

The issue isn't capability. Tools like the Intuit/Anthropic integration in QuickBooks, Microsoft 365 Copilot at $21/month, and Fathom for meeting summaries (free tier available) are genuinely useful for accounting workflows right now. The issue is that access to a tool does not create a habit, and a habit does not create a workflow, and a workflow does not create firm-wide consistency without deliberate structure.

That structure is what most firms are missing.

Step 1: Start Small — One Workflow, One Team, One Month

The most common adoption mistake is trying to solve AI "broadly" before solving it specifically.

Firm owners who announce "we're going to start using AI" without naming a specific workflow, a specific team, and a specific timeline are not setting up an adoption plan. They're setting up a suggestion box. The initiative will produce a handful of experiments and then quietly fade.

The alternative: pick one pain point. Not "client communication" — that's too broad. Pick "draft follow-up emails after client calls using Fathom meeting summaries." Or "use AI to generate the first draft of an engagement letter from a template." Or "use Copilot to draft responses to routine client questions in Outlook."

One workflow. For one team. For one month. Then measure what changed — time saved, quality difference, team comfort level.

Once the first workflow is working, it's easy to add a second. Before the first one is working, adding a second creates confusion and dilutes adoption energy.

What to actually start with this week: Meeting summaries are the fastest path to daily AI use for most accounting teams. Fathom integrates with Zoom and Teams, records and transcribes calls automatically, and generates structured summaries in 60 seconds after the call ends. Setup takes 15 minutes. No client data enters a third-party system — just the audio of the call. The whole team can see the value of AI within the first client meeting they run with Fathom active.

Step 2: Invest in Learning — the Minimum That Actually Works

Only 17% of workers globally believe their employers are investing in the skills they need for AI. That gap is the single biggest predictor of whether staff will adopt AI or quietly ignore it.

"Invest in learning" does not mean buying a course or sending a link to documentation. Most of those interventions produce nothing.

The minimum effective learning investment for a small accounting firm:

One 90-minute hands-on session per team. Not a demo. A session where every person opens the tool, runs a real example with actual work from their queue, and asks questions in the room. The person running the session should be the owner or a senior staff member — not an outside consultant, not a recording.
One designated AI lead. After the session, name one person whose informal job it is to stay current on how the firm uses AI, answer questions, and flag what's working. In a 10-person firm, this is typically a 20-30 minute per week informal responsibility. It does not require a new title or additional compensation — just acknowledgment that someone is responsible.
A 30-day check-in. One month after the initial session, spend 15 minutes in a team meeting on one question: what has worked, what hasn't, what should we stop doing or keep doing? This closes the feedback loop and signals that AI adoption is an ongoing commitment, not a one-time announcement.

That's it. Three things. Firms that do these three things see adoption hold. Firms that skip them see adoption revert within 60 days.

Step 3: Communicate the Why — Before the What

Here's what happens in most firms when AI tools get introduced: the owner sends a Slack message or email announcing the new tool, maybe includes a link to the vendor's help documentation, and assumes the team will figure it out.

What the team hears: "We're replacing something you do with software."

That gap between intent and reception is why staff resistance to AI tools is so common — even when the tools are genuinely good. People don't resist AI because they're Luddites. They resist it because they don't know what it means for their job security, their professional development, or whether their supervisor trusts them to use it correctly.

Andrea Wynter's framework is explicit on this point: communicate why the tools are being introduced before you communicate what they are. Specifically:

Explain the purpose. Is this about reducing time on low-value tasks so staff can do more complex work? About allowing the firm to take on more clients without adding headcount? About reducing errors in routine documentation? Name it. Don't make staff guess.
Establish an ethics policy. Which client data can go into AI tools under what circumstances? What requires human review before going to a client? What AI-generated output should never be used without a professional review step? Write it down. Make it one page. Give it to everyone.
Frame AI as augmentation, not replacement. For most firms, this is genuinely true — AI is handling the volume work so humans can do the judgment work. If you mean this, say it. If it's not true for every role, say that too, clearly and directly. Dishonesty in either direction destroys trust faster than any technology change.

Step 4: Maintain Human Oversight — for Professional and Liability Reasons

The final step is not optional for licensed professionals.

Every AI output that reaches a client or regulatory authority must pass through a professional review step. This is an ethical obligation under CPA licensing standards, increasingly a regulatory requirement at the state level, and simply good practice given that AI systems produce plausible-sounding but occasionally incorrect outputs.

In practical terms, this means:

AI output is always a first draft. A licensed professional reviews it before delivery.
Complex tax situations, novel fact patterns, and judgment calls do not get AI-first treatment. They get professional-first treatment, with AI potentially assisting in supporting research or documentation.
Routine tasks — summarizing a client call, drafting a standard engagement letter, generating a first-pass variance explanation — are the appropriate AI use cases where oversight is a check step, not a full rework.

The good news: building oversight into your workflow doesn't slow you down if the workflow is designed correctly. The firms that try to use AI without a review step — to save the most time — are the ones who will eventually send something to a client that shouldn't have been sent. Build the review step in from the beginning.

The Real Question Is Not "Which Tool?"

The 1-in-5 daily adoption rate exists because most accounting firm owners are asking the wrong question.

They're asking: which AI tool should we buy?

The right question is: how do we build a workflow where AI use is the path of least resistance for our team's most time-consuming tasks?

That question has a different answer for every firm. But the four steps above give you a structure to find it. Start with one workflow. Run one real training session. Explain the why before the what. Build in oversight from the beginning.

The firms that are in the 19% didn't get there because they found a better tool. They got there because they built the structure around the tools they had.

One Specific Thing to Do This Week

If your firm isn't using AI daily, pick one person on your team — ideally someone who expressed curiosity about AI tools — and spend 30 minutes this week running their most time-consuming repetitive task through a free AI tool together.

Meeting summaries: Fathom (free for individuals, team plans from $19/month). Client email drafts from notes: Claude.ai or ChatGPT (free tier, browser-based). Excel formula writing and data transformation: Microsoft 365 Copilot ($21/user/month if you're on M365).

Don't announce a firm-wide initiative. Don't schedule a training session yet. Just run one task through one tool with one person, and see what you both think.

That's how the 19% started.

Source: CPA Practice Advisor, March 10, 2026 — "How Accounting Firms Can Close the AI Adoption Gap". ADP research; framework attribution to Andrea Wynter, VP, ADP Canada.

Only 1 in 5 Accountants Uses AI Every Day — Here's the 4-Step Framework to Change That at Your Firm