Artificial intelligence is forcing tax and accounting firms to confront a much bigger question than which tools they should buy.
What will the firm of the future actually look like?
Right now, three possible visions are emerging.
The first is the Silicon Valley vision. AI-native platforms will rebuild the firm from the ground up, automate most of the work and disrupt the legacy providers that have served the profession for decades.
The second is the legacy software vision. Firms will continue using the same established platforms and workflows while those providers add AI drafting, search, summarization and automation features around the edges.
The third possibility is somewhere in between.
AI will create entirely new capabilities, but professional judgment, client relationships and human accountability will remain central. The technology will not simply replace the firm. It will make the professionals inside it far more capable.
That third path is the one I believe is most likely to win.
But it also creates a harder question for firm leaders:
How do you know whether your AI investment is actually working?
Three Competing Visions for the AI-Enabled Firm
Silicon Valley looks at tax and accounting and sees an industry ready for disruption.
The software is often fragmented. Many workflows remain manual. Information is trapped across tax systems, accounting platforms, payroll records, inboxes, workpapers and client portals.
Professionals spend too much time gathering information, moving data and remembering what should happen next.
From the outside, the answer seems obvious.
Replace the old systems. Rebuild the workflows around AI. Deploy autonomous agents. Remove as much human labor as possible.
There is real logic behind that vision.
Some existing workflows are inefficient. Some legacy platforms were designed for a very different era. Simply adding a chatbot or a few drafting tools to outdated software will not necessarily create a modern operating model.
But tax and accounting firms do not operate inside software demonstrations.
They have existing clients, trained employees, filing deadlines, regulatory obligations and processes developed over many years. Their work involves incomplete facts, changing laws, ethics, risk tolerance and professional responsibility.
The person signing the return or delivering the recommendation still owns the result.
That makes replacing everything overnight much more difficult than the disruption narrative suggests.
At the opposite end are the legacy technology companies.
Many are adding useful AI features to systems firms already understand. Those features may help draft emails, summarize documents, locate information or automate routine tasks.
For some firms, that may be enough for now.
Accountants are naturally cautious about disrupting processes that have produced reliable results. If employees understand the workflow, clients receive consistent service and the work gets completed accurately, there may be no reason to replace the entire process immediately.
But legacy providers face a different risk.
They can add AI features without changing what the firm is actually capable of doing.
The same people may still initiate every process.
The same partners may remain bottlenecks.
The same institutional knowledge may remain trapped inside the heads of a few experienced professionals.
The same client opportunities may continue to be missed.
The firm becomes faster, but not necessarily more intelligent.
That is why the most likely future sits between the two extremes.
AI Should Make the Firm More Capable
The most important question is not whether AI can complete a task faster.
It is whether AI makes the firm more capable.
Can the firm recognize client needs earlier?
Can it prepare professionals more effectively?
Can it make its best internal knowledge available across the organization?
Can it deliver more consistent work?
Can it create capacity without requiring the same increase in labor?
Can it improve margins, client retention and employee experience?
Those outcomes matter far more than the number of prompts saved or licenses purchased.
AI success should be measured by what changes inside the business.
Time Saved Is Only the Starting Point
Most firms begin by measuring speed.
An email that once took 20 minutes now takes two.
A meeting summary is created instantly.
A document review that once took an hour now takes 15 minutes.
Those improvements matter, but they are not the return.
Suppose AI saves a firm 100 hours a month.
If none of that time is redirected into faster delivery, more client work, advisory capacity or better service, the firm has created theoretical capacity, not captured value.
That is the distinction.
What happened to the time that was saved?
Did the employee complete more client work?
Did turnaround time improve?
Did the firm reduce a backlog?
Did a professional have more time for a client conversation?
Did the firm create room for advisory services?
Or did the time disappear into more email, meetings and administration?
Time saved creates potential capacity.
The firm still has to capture and direct it.
That is why AI ROI cannot be calculated only by multiplying minutes saved by an hourly rate.
The more important question is what the firm became capable of doing with the time.
Four Ways to Measure AI Success
A practical AI scorecard for tax and accounting firms should focus on four areas.
1. Adoption and Professional Trust
The first measure is whether professionals actually use and trust the technology.
Not whether they logged in.
Not whether they attended training.
Not whether they experimented with it once.
Is AI becoming part of the way real client work gets completed?
This matters because theoretical capability has little value if professionals avoid the system.
A highly advanced AI platform that employees do not trust may create less value than a simpler tool they use every day.
Firm leaders should ask:
- Which AI workflows are being used repeatedly?
- Where are employees correcting the output?
- Where are they abandoning the AI process and returning to the old method?
- Do professionals understand the sources behind the result?
- Do they know when human review is required?
- Does the system fit into the actual workflow?
Professional caution should not automatically be dismissed as resistance to change.
Tax and accounting professionals remain responsible for the work they deliver. Wanting to understand how a conclusion was reached, which information was used and where the risks remain is often appropriate professional judgment.
At the same time, caution cannot become an excuse to preserve every old process forever.
Trust must be earned through reliability, transparency, useful context and clear boundaries.
Adoption is not a soft measurement.
It is part of the ROI.
2. Capacity and Consistency
The second measure is whether AI creates usable capacity and makes the work more consistent.
One of the largest opportunities is reducing the cost of preparation.
Professionals spend enormous amounts of time locating documents, searching through emails, reviewing prior notes and reconstructing client history.
Only after gathering the information do they reach the work the client is actually paying for.
Judgment.
AI should help gather, organize, compare and summarize the relevant information before the professional begins the analysis.
That allows the professional to spend less time assembling facts and more time interpreting them.
The goal is not to remove professional judgment.
It is to stop wasting professional judgment on work technology can perform.
AI can also help firms make their best internal processes more repeatable.
Every firm has experienced professionals who know which questions to ask, which facts matter and which issues are easy to overlook.
Much of that knowledge exists inside conversations, review notes and individual memory.
A well-designed AI skill can help capture the process.
It can define the information required, the questions that should be asked, the sources that should be reviewed, the output that should be produced and the points requiring professional approval.
The AI does not replace the experienced professional.
It helps more employees begin with a better process.
The real measurement is whether the firm’s expertise becomes easier to repeat and less dependent on one person remembering everything.
3. Client Intelligence and Action
The third measure may represent the largest opportunity for tax and accounting firms.
Did AI help the firm turn client information into action?
Most firms already possess an enormous amount of useful data.
It sits inside tax returns, accounting systems, payroll records, emails, meeting notes, uploaded documents and prior-year workpapers.
The problem is that much of that information remains inactive.
A client approaches retirement.
Nobody notices.
A business begins operating in several states.
Nobody connects the information.
An S corporation reports significant distributions and relatively low wages.
The return is completed, but the issue never becomes a proactive conversation.
A client’s income changes in a way that may affect student-loan payments.
The planning opportunity is missed.
No professional can continuously review every piece of information across hundreds or thousands of clients.
AI can help the firm recognize what it already knows.
That is the difference between productivity and intelligence.
The productivity question is:
Can AI write an article about Social Security?
The intelligence question is:
Which clients may need a Social Security planning conversation?
The first creates content.
The second creates an opportunity to help a client.
AI might identify relevant clients, organize the known facts, highlight missing information and prepare an initial briefing.
The professional still reviews the information, determines whether the issue applies and decides how to proceed.
The technology does not create the advice.
It helps identify where professional attention may be needed.
But intelligence alone is not enough.
The firm still needs a process for turning the signal into action.
Who reviews the opportunity?
Who contacts the client?
What service is offered?
How is it priced?
How is the outcome tracked?
The return appears when the process becomes:
Opportunity identified.
Professional review.
Client conversation.
Proposal.
Engagement.
Delivery.
Follow-up.
That is how AI becomes a growth engine rather than another report.
4. Business and Human Outcomes
The fourth measure is whether AI improves both the economics of the firm and the experience of the people inside it.
From a business perspective, firms should evaluate whether AI contributes to:
- Revenue growth
- Stronger margins
- Better retention
- Faster lead response
- More advisory opportunities
- Greater operating leverage
- More efficient service delivery
Revenue alone is not enough.
AI may help identify additional advisory work, but the economics will remain poor if every engagement still requires excessive preparation, partner involvement and rework.
The firm should compare the cost of delivering the service before and after AI.
How much staff time was required?
How much manager review?
How much partner involvement?
How many client follow-ups?
How much rework occurred?
How long did the process take?
If AI allows the firm to deliver a stronger result more efficiently, the firm should capture some of that value through better margins.
Clients are not buying the number of hours required to produce the answer.
They are buying the result.
The human outcomes matter just as much.
Are partners spending more time exercising judgment and less time searching for documents?
Are managers spending more time coaching and less time correcting preventable inconsistencies?
Are employees developing higher-value skills instead of copying information between systems?
Are clients receiving faster, more proactive and better-prepared service?
The goal is not to automate the relationship.
It is to automate the work that keeps the relationship from happening.
Traditional Workflows Still Have a Role
Firms do not need to replace every traditional workflow to benefit from AI.
Traditional workflows remain valuable for predictable processes.
They tell the firm whether documents have been requested, whether work is in preparation, whether it is awaiting review or whether the client needs to respond.
That structure provides visibility and accountability.
But a traditional workflow primarily tells the firm where the work is.
An AI workflow can help the firm understand what the work requires.
Imagine a new business client entering the firm.
A traditional workflow may create an onboarding project, send an intake form and request documents.
An AI-enhanced workflow could review the intake and recognize that the company has employees in three states, the owner appears to have low S corporation wages and the business has no retirement plan.
It could identify missing information and prepare questions for the advisor.
The professional still reviews the facts and makes the decision.
But the advisor enters the meeting better prepared.
The traditional workflow tells the firm where the client is in the process.
The AI workflow helps the firm understand what the client may need.
That is the practical path most firms should pursue.
Not replacing everything immediately.
Not preserving everything indefinitely.
Using intelligence where it can improve the outcome.
Professional Judgment Remains the Boundary
Every measure of AI success must sit beneath one overriding requirement.
The firm must preserve quality, security, judgment and professional accountability.
A process is not successful merely because it is faster, more profitable or more widely used.
The result still has to be reliable.
The firm still owns the final output.
AI can help identify issues, organize facts, frame questions, summarize information and challenge an initial conclusion.
But it does not eliminate due diligence.
The professional still needs to verify the authority, determine whether it applies and decide whether the client’s facts support the conclusion.
Human review is not evidence that AI failed.
In professional services, it is often evidence that the process was designed correctly.
Silicon Valley sometimes treats the human in the loop as a temporary inconvenience that technology will eventually remove.
In tax and accounting, the human in the loop is often the product.
The client is paying for interpretation, judgment, accountability and the willingness to stand behind a recommendation.
The goal is not to automate judgment.
It is to give professionals more time and better information to exercise it.
The Real Measure of AI Success
The future of AI in tax and accounting will probably not belong entirely to AI-native disruptors.
It will probably not belong entirely to legacy providers adding isolated AI features either.
The firms that win will build a bridge between technological capability and professional reality.
They will preserve the processes that still work and replace those that no longer make sense.
They will use AI to reduce preparation, connect information and make their best thinking repeatable.
They will introduce new capabilities in ways professionals can understand and trust.
They will turn client information into proactive service.
They will automate routine work without automating accountability.
Most importantly, they will measure AI based on what changes inside the business.
When evaluating an AI investment, firms should ask four questions:
- Are our professionals using and trusting it inside real work?
- Is it creating usable capacity and making our expertise more repeatable?
- Is it turning client information into intelligence and action?
- Is it improving business results and the experience of our clients and employees?
Then one requirement should sit across all four:
Are we preserving quality, security, professional judgment and accountability?
For the last several years, everyone has been asking:
What can AI do?
The more important question now is:
What should our firm become capable of doing because AI exists?
That is what meaningful AI success looks like.
‍
