What's one of the easiest ways for an accounting firm to start using AI?
It may not be asking AI to do accounting at all.
Instead, use AI to help your accountants and administrative teams organize information, identify what's missing, prepare for client conversations, and ask better questions.
For accounting firms exploring tools like Microsoft Copilot or ChatGPT, this is a practical place to start, because it supports the work without replacing the professional judgment behind it.
Anyone in accounting knows the scenario. A client sends an email with half the information you need. There are three attachments. One document is missing. A number doesn't quite add up. And somewhere in a long email thread is the one detail you actually need before you can move forward.
Now multiply that by dozens — or hundreds — of clients.
Especially during tax season, year-end, or other busy periods, a surprising amount of time gets spent figuring out what's missing, following up, organizing responses, and following up again when only half the request gets answered.
Your next job isn't necessarily to solve the problem. It's figuring out what you still need from the client. And that's something AI can help with.
When firms first experiment with AI, there's a natural pull toward big questions:
That's not where an accounting firm should start.
Instead, give AI a narrower job: organize, prepare, identify, or clarify.
Rather than asking:
"How should I account for this?"
Try:
"What information appears to be missing before I can properly review this?"
That's a very different role for AI. The accountant stays responsible for determining what's relevant and applying professional expertise — AI simply helps create a starting point.
Here's a prompt your team can test with a non-sensitive example:
I'm going to describe a client request or situation. Based only on the information I provide, identify what information appears to be missing or unclear.
Create a concise list of questions I could send to the client to gather the information needed to move forward.
Organize the response into:
- Information already provided
- Information that appears to be missing
- Questions I should ask the client
- Documents or records I may need to request
Do not make assumptions or invent missing information. Do not provide tax, accounting, or legal advice. If you are unsure whether something is required, flag it for professional review.
Finally, turn the questions into a friendly, professional email I can review before sending.
That last instruction is what turns this into a real time-saver. Instead of starting a client follow-up email from a blank page, the accountant gets an organized first draft to review, correct, and personalize.
Imagine a client sends this message:
"We bought a new vehicle for the business earlier this year. Some of it was paid personally and some through the company. I've attached the invoice. Can you make sure it gets included with everything this year?"
You could ask AI what to do with the purchase — but AI won't have enough information, and you'd be pushing it into territory that requires real professional judgment.
Instead, run the situation through the Client Question Builder. AI might surface questions like:
Here's the important part: the accountant reviews the questions. Some will be useful. Some won't apply. An experienced accountant will likely spot something AI missed entirely.
That's exactly how this is supposed to work: AI creates the starting point. The professional provides the judgment.
The same approach extends to plenty of the administrative work that surrounds accounting itself.
Give AI sanitized meeting notes and ask:
Based only on these notes, create a meeting agenda and identify the questions that should be answered before the meeting ends. Clearly identify anything that requires professional review.
This turns scattered notes into a more structured conversation before you ever sit down.
Say your notes look something like:
"Client mentioned purchasing equipment in June. Think there may be financing. Need invoice. Also mentioned selling old equipment but didn't provide details. Hiring someone in September. Need to follow up."
Ask AI:
Turn these notes into a client follow-up checklist. Separate confirmed information from information that still needs clarification. Do not assume anything that isn't included in my notes.
Then follow with:
Using the checklist, draft a concise, friendly client follow-up email. Make each request easy for the client to answer.
Now you have a starting point instead of a blank email.
AI can also review your own draft. Try:
Review these questions from the client's perspective. Identify anything that is repetitive or unclear and rewrite it so the client understands exactly what information or documentation I'm requesting. Do not change the meaning of the questions.
This can meaningfully cut down on unnecessary back-and-forth.
AI can also take non-sensitive notes from an internal meeting and organize them into:
Again — AI isn't doing the accounting. It's helping organize the work around it.
If you're trying to figure out where AI fits into your firm, don't start with your most complicated processes. Start with the small, repetitive ones — the five-minute tasks.
Think about how often someone on your team:
Saving five minutes once isn't exciting. Saving five minutes on something your team does every week can be.
More importantly, these lower-risk tasks build an important habit:
Prompt → Review → Question → Correct → Use
That's a much healthier pattern than:
Prompt → Copy → Send
Potentially — and this may be one of the more interesting opportunities for professional services firms.
Compare asking a client:
"Can you send us more information about the vehicle?"
with sending a short, specific list of exactly what you need.
If AI helps you organize that initial request more clearly, it can make things easier for both sides. After creating your questions, try this follow-up prompt:
Review the questions you created. Can any be combined or rewritten so the client receives the shortest, clearest request possible without losing important information?
Then review the result yourself. The goal isn't to have AI communicate autonomously with your clients — it's to help your team create clearer communication, more efficiently.
A common question from firms exploring AI: should our employees use Microsoft Copilot, ChatGPT, or something else?
There's no single answer that fits every accounting firm. The more important questions are:
The right AI conversation isn't only about features. It's also about data governance, cybersecurity, access, and employee training. Before rolling out any AI platform firm-wide, understand how it fits into your existing technology and security environment.
This matters more for accounting firms than almost any other industry. Accounting professionals routinely handle highly sensitive information, including:
That doesn't mean accounting firms shouldn't use AI — it means they need clear rules around how it's used.
Before an employee enters business information into an AI platform, they should know which tools are approved and what information is permitted to be entered.
For a training exercise like today's prompt, strip out identifying information wherever possible. Instead of:
"John Smith from ABC Inc. purchased a $67,432 vehicle..."
Use:
"A business client purchased a vehicle during the year..."
You can learn the technology without unnecessarily exposing client information.
As employees start using AI regularly, it's worth revisiting your firm's Acceptable Use Policy and AI guidelines. Employees should understand:
Blocking every AI tool isn't an AI strategy. Neither is letting everyone use whatever they want. The goal is useful guardrails.
There's one more prompt worth adding to your toolkit:
What are you assuming?
After AI generates an answer, follow up with:
Review your response and identify anything you assumed that wasn't explicitly included in the information I provided.
This is a good habit regardless of which AI platform you use. AI can produce an answer that's polished, organized, and completely confident — and that doesn't mean it's correct. Learning to challenge the output instead of automatically accepting it is one of the most important parts of getting comfortable with AI.
For accounting firms, some of the best early AI opportunities are the least dramatic ones. Use AI to help your team:
Then let the accountant do what the accountant does best: apply professional knowledge, context, experience, and judgment.
Think of one client conversation that still needs a follow-up. Strip out any sensitive or identifying information and describe the situation to your approved AI tool. Then ask:
"What am I missing, and what should I ask next?"
Review its suggestions. Challenge its assumptions. Remove anything irrelevant. Add anything your professional experience tells you is missing. Then ask AI to turn your final list into a clear client follow-up.
You haven't asked AI to be the accountant. You've asked AI to help the accountant get organized. And that's a great place to start building AI confidence.
AI adoption shouldn't begin and end with buying a Copilot licence. Your technology, Microsoft 365 environment, permissions, cybersecurity, employee training, and policies all shape how safely and effectively your team can use AI.
Attitude IT helps Ontario businesses explore practical ways to introduce AI while keeping cybersecurity and responsible use at the centre of the conversation.
If your accounting firm is wondering where to start with Microsoft Copilot, AI policies, cybersecurity, or employee AI training, let's have a conversation. Check out 30 days to AI-Confidence at www.attitudeit.ca/ai-confidence and other articles on IT Support at www.attitudeit.ca.