Blog
AI for Small Accounting Firms: What to Use AI For, and What to Automate First
Casia, Rosada Automation11 min read
AI helps a small accounting firm most with language work: drafting client emails, summarizing intake answers and long threads, and sorting incoming requests, always with a person reviewing before anything goes out. The repeat, rule-based work around it, like engagement-letter reminders, document chasing, the "ready for prep" handoff, and invoice reminders, is better handled by plain automation, which does the same thing every time. Start with one workflow that slips every season, automate the rules, and add AI only where someone is currently writing or reading the same thing over and over.
Not sure where AI would actually save your team time? Get your free workflow map. About 2 minutes. Free, no call.
AI or automation? What's the one-line test?
AI is for work that involves reading or writing language and benefits from a first draft: a reply, a summary, a sort. Plain automation is for work that follows a rule: if this happens, do that, and stop when something else happens.
Here's the test I use: would two people on your team do this exactly the same way every time? If yes, it's a rule, so automate it. If the wording changes with each client, AI can draft it and a person approves it.
The AICPA & CIMA's small-firm advice points the same way: map a typical engagement from intake to delivery, and look for manual, repetitive, or information-heavy tasks. As Nicole Davis, owner of Conscious Accounting, says in that article, "fix the problem that you have." Find the logjam first, then pick the tool.
The AI inside your ledger and tax software is the accounting side. This post is about the firm's admin around the returns.
How does AI fit into the five workflows a small firm already has?
Most small firms run the same five client workflows:
| Workflow | Plain automation does | AI can help with | A person always |
|---|---|---|---|
| 1. New inquiry | Same-day reply, intake form, booking link, record saved | A short summary of the intake answers; sorting by service type | Takes the call, decides fit |
| 2. Engagement letter | E-sign request after approval, scheduled reminders, stops on signature | A first draft of the scope paragraph from call notes | Approves every letter |
| 3. Document requests | Item-level request; reminders list only what's missing | A plain-English note on what a requested item is, from a template you approved | Reviews uploads |
| 4. Ready for prep | Status change, preparer notified, client told "we have everything" | No AI needed | Does the prep |
| 5. Unpaid invoices | Friendly reminders that stop on payment | No AI needed for routine reminders | Disputes and payment arrangements |
New inquiry. The reply and booking link are rules. AI earns its keep with a short summary before the call, so nobody reads a long form cold. The Journal of Accountancy describes Agate CPA, a six-person firm, rebuilding its prospect intake form with AI tools so answers flow straight into its practice management system. See the call notes to CRM sample run (sample data) for the "AI drafts, you check" pattern.
Engagement letter. AI drafts the scope paragraph from call notes; a person approves every letter. Nothing goes out under the firm's name unchecked.
Document requests. AI can explain a requested item in plain English, from wording you approved. A person reviews uploads. No AI decides a file is complete.
Ready for prep. All rules. No AI here.
Unpaid invoices. Routine reminders are rules. For a sensitive overdue note, AI can draft and someone reviews. See the unpaid invoice reminders sample run (sample data).
Two of the five don't need AI at all. That's normal.
The step-by-step for each is in what to automate in a small accounting firm.
Prompts and messages you can borrow
Intake summary prompt (AI, in a firm-approved tool): Summarize these intake answers in five bullets for the partner taking the call: services asked about, entity type, deadlines mentioned, open questions. Don't guess anything that isn't in the answers.
Scope draft prompt (AI): From these call notes, draft a one-paragraph scope of services in plain English. Mark anything uncertain as [CHECK].
Unsigned engagement letter (automation, no AI): Hi [First name], a quick reminder that your engagement letter from [Firm] is waiting for your signature: [secure link]. Questions? Just reply.
Missing documents (automation, no AI): Hi [First name], thanks for what you've uploaded. We're still missing: [item list]. You can add them here: [portal link].
Which one should you set up first?
Rate each one in words, not numbers:
| Workflow | How often | Cost when it slips | Risk if it's wrong | Needs AI? |
|---|---|---|---|---|
| New inquiry | Medium | Medium | Low | Optional (the summary) |
| Engagement letter | High in season | High | Low | Optional (the scope draft) |
| Document requests | High in season | High | Low to medium | No |
| Ready for prep | High in season | Medium | Low | No |
| Unpaid invoices | Medium | Medium | Medium | No |
I'd start with engagement-letter reminders or document requests: they repeat every season, are low-risk, and need no AI. Add the AI intake summary second.
The workflow, start to finish: new client to ready for prep, with AI marked
The dashed steps are where AI drafts and a person reads or approves. Everything else is a rule with a clear stop.
The example workflow at a glance: AI drafts the writing, automation runs the rules, and your team keeps the judgment.
Want this mapped for your firm, with the AI steps marked? The free workflow map shows how a new client moves from inquiry to ready-for-prep today, and where it stalls. About 2 minutes. Free, no call.
What are the best AI tools for accounting firms?
Pick by job, not by brand:
- AI already built into the software you pay for. The AICPA notes that many AI and automation tools are already embedded in software small firms use every day. Check what's in your plan before buying anything new.
- A general AI assistant on a business plan your firm has approved. For drafting, summarizing, and internal procedures.
- A research tool built on authoritative sources. For technical questions. The AICPA describes Josi, its generative AI research tool, as giving CPAs access to authoritative AICPA professional materials.
- An automation platform that connects your tools. Form to CRM to e-sign to portal. The AICPA & CIMA's small-firm automation guide names Zapier, Make, and Microsoft Power Automate as ways to do this without programming skills.
The AICPA's advice for picking: start small, test one tool on one workflow, and scale only what works. If you'd rather have someone build it, here are the questions to ask before you hire anyone.
What rules already apply to client data and AI?
This is general information, not tax, legal, or security advice.
- The IRS has introductory guidelines. In June 2026 the IRS Office of Professional Responsibility issued Introductory Guidelines for Responsible AI Use in Federal Tax Practice. It applies existing Circular 230 duties to AI rather than creating a new law. In short: review all AI-created documents before they go to a client or the IRS, verify facts, citations, and calculations, and don't rely on AI alone. Handle client data only in secure, enterprise-approved AI, and never upload sensitive data to unsecured sites. Firms should also have AI policies, staff training, and vetting for third-party AI tools.
- Your security plan should cover these tools. IRS Publication 4557 says the FTC Safeguards Rule covers professional tax preparers, who must create and enact security plans to protect client data. Publication 5708, the IRS's sample written information security plan (WISP), has you document where client information is stored, including cloud apps, and check that third-party providers keep safeguards. Every AI and automation tool that touches client data belongs on that list.
- There's a policy starting point. The AICPA publishes a Small Firm Generative AI Policy Template for members.
- My own rule: no SSNs or full tax documents in a general chatbot, ever.
What not to hand to AI
- Tax positions, return reviews, and advice that needs professional judgment
- Anything signed or sent under the firm's name that a person hasn't checked
- Any step that puts client documents or ID numbers into a tool your firm hasn't approved
When don't you need AI at all?
If you're a solo practice with a handful of clients, or your portal already sends every reminder and your team uses it, plain automation, or nothing new, is enough.
Start with a map
See a new client onboarding example run (sample data). This is the kind of thing I build, in your accounts.
FAQ
How can a small accounting firm use AI?
Mostly for language work, reviewed by a person: drafting client emails, summarizing intake answers and long threads, and sorting incoming requests. The repeat steps around client work, like engagement-letter reminders, document chasing, and invoice reminders, are better handled by plain automation. Tax positions, reviews, and advice stay with your team.
Should an accounting firm start with AI or with automation?
Usually automation. Engagement-letter reminders and document requests repeat every season, follow clear rules, and are low-risk, so they don't need AI. Once one of those runs reliably, add AI where someone keeps writing or reading the same thing, such as summarizing intake answers before a first call.
What are the best AI tools for accounting firms?
It depends on the job. Check the AI already built into the software you pay for first. Then consider a general AI assistant on an approved business plan for drafting, a research tool built on authoritative sources for technical questions, and an automation platform to connect your tools. Test one tool on one workflow before buying more.
Is it safe to put client information into AI tools?
Only in secure, firm-approved tools that are listed in your written information security plan. The IRS Office of Professional Responsibility's 2026 AI guidelines say to protect client confidentiality, vet third-party AI tools, and review everything AI produces. Don't put SSNs or full tax documents into a general chatbot. This is general information, not legal advice.
Does the IRS have guidance on AI use by tax professionals?
Yes. In June 2026 the IRS Office of Professional Responsibility issued introductory guidelines for responsible AI use in federal tax practice. They say practitioners must review AI-drafted work, verify facts, citations, and calculations, keep client data secure, and have firm policies and training for AI use.
What shouldn't an accounting firm use AI for?
Tax positions, return reviews, and advice that needs professional judgment, plus anything sent under the firm's name that a person hasn't checked. Also skip AI where a simple rule works, like routine reminders and status updates. A rule does the same thing every time and is easier to check.
Sources
- IRS Office of Professional Responsibility, "Issue Number 2026-19: Introductory Guidelines for Responsible AI Use in Federal Tax Practice" (June 24, 2026): https://www.irs.gov/pub/opr-taxpros/issue-number-2026-19-introductory-guidelines-for-responsible-ai-use-in-federal-tax-practice.pdf
- AICPA & CIMA, "Beyond Email: Using AI and automation to enhance capacity" (Feb 11, 2026): https://www.aicpa-cima.com/resources/article/beyond-email-using-ai-and-automation-to-enhance-capacity
- AICPA & CIMA, "How to build automations in your small firm" (Dec 3, 2024): https://www.aicpa-cima.com/professional-insights/article/how-to-build-automations-in-your-small-firm
- Journal of Accountancy, "Real-life ways small firms use AI" (Aug 2026): https://www.journalofaccountancy.com/issues/2026/aug/real-life-ways-small-firms-use-ai/
- AICPA & CIMA, "AICPA Small Firm Generative Artificial Intelligence (AI) Policy Template" (Apr 29, 2026): https://www.aicpa-cima.com/resources/download/aicpa-small-firm-generative-artificial-intelligence-ai-policy-template
- IRS Publication 4557, "Safeguarding Taxpayer Data": https://www.irs.gov/pub/irs-pdf/p4557.pdf
- IRS Publication 5708, "Creating a Written Information Security Plan for your Tax & Accounting Practice": https://www.irs.gov/pub/irs-pdf/p5708.pdf
- Rosada Automation, "Call notes to CRM" (example run with sample data): https://rosadaautomation.com/automations/call-notes-to-crm
- Rosada Automation, "Unpaid invoice reminders" (example run with sample data): https://rosadaautomation.com/automations/unpaid-invoice-reminders
- Rosada Automation, "New client onboarding" (example run with sample data): https://rosadaautomation.com/automations/new-client-onboarding
Casia