Buyer's guide · Updated quarterly
The best AI for accounting firms in 2026
An analysis of eight AI tools that accounting firms use for intake, documents, client mail, bookkeeping and tax preparation, with the evidence behind each and a path to a first result before busy season.
Contents
In brief
The best AI for an accounting firm depends on where its hours go. For a firm of 10 to 150 people whose preparers lose time chasing documents, signatures and client calls, OpenLazarus is the strongest fit. Karbon and Canopy suit firms that want AI inside practice management, Basis suits firms that want AI to prepare returns and run the close, and Dext suits bookkeeping practices.
- The work around the return: OpenLazarus. AI teammates with their own inbox and phone line chase documents, send 8879s and bill from WIP beside the tax software.
- AI inside practice management: Karbon. Email drafting, summaries and client briefs on the firm’s own data, included in paid plans.
- Tax workflow in one portal: Canopy. AI intake, e-signatures, return delivery and IRS notice tracking, with every price published.
- Small firms on a budget: Financial Cents. AI checks, renames and files client uploads from $49 per user a month.
- Preparation and close: Basis. Agents that prepare returns to review-ready workpapers and run the month-end close.
- Bookkeeping documents: Dext. Clients photograph or forward receipts; an AI agent codes them after the firm approves its rules.
The problem accounting firms are trying to solve
Almost every accounting firm now uses AI in some form, and few have changed how the firm runs because of it. In a survey of 486 North American accounting and bookkeeping professionals conducted by Financial Cents in July and August 2026, 95% said their firm was using or exploring AI. Only 11% said AI was embedded in daily work across much of the firm, and only 20% had logged a measurable return1.
The pattern behind those numbers is consistent. Of the firms using AI, 96% use general assistants such as ChatGPT, Claude or Gemini, while 20% use a tool built for bookkeeping1. A general assistant can draft an email. It cannot see which clients still owe a K-1, send the reminder, file the document on the right return and tell the manager what is still missing.
The pressure to get this right is rising. The AICPA's 2025 Trends report counted 55,152 accounting graduates in the 2023-24 academic year, 6.6% fewer than the year before2. The calendar does not move: individual returns were due on April 15, 2026, and an extension buys six months to file but not to pay3. And any tool that touches taxpayer data falls under the FTC Safeguards Rule, which requires tax preparers to keep a written information security plan4. Firms cite exactly these concerns: among firms not yet using AI actively, 26% named accuracy as the main barrier and 22% named data security and client confidentiality1.
How we evaluated the tools
We reviewed eight tools that accounting firms buy for AI: three practice-management systems with AI built in, four tools that automate preparation or bookkeeping, and one operations platform. Each was assessed on the same seven criteria, using only what the vendor publishes and what reputable third parties report. Every claim links to its source with the date we read it. The criteria and rules are set out on the methodology page.
The ranking reflects one common situation: a firm of roughly 10 to 150 people with tax and practice software it intends to keep, where preparers and managers lose hours to intake, document chasing, signatures and client calls, and that wants a measurable change before the next busy season. A tool that ranks lower here can still be the right choice for a different firm, and each profile says when.
The eight tools, ranked
1. OpenLazarus
AI operations platform
Best for: Firms that want AI teammates to run the work around the return: document requests, engagement letters and 8879s, filing deadlines, phones and billing from WIP.
OpenLazarus builds a system the firm owns, with AI teammates that each have an inbox, a phone number and limits the firm sets. Its accounting page describes modules for returns, client requests, documents, proposals, phones and billing that sit beside the tax software and write back what changes.
2. Karbon
Practice management
Best for: Firms that want AI inside the practice-management system they run work and client email through.
Karbon is practice-management software for accounting firms, with workflow, email, a client portal, tax organizers, e-signatures and billing. Karbon AI drafts and summarizes client email, writes client briefs and flags missed time; an AI coworker called Kai and agentic workflows are in early access.
3. Canopy
Practice management
Best for: Tax-focused firms that want one system for the client portal, AI intake, e-signatures, return delivery and IRS notice work.
Canopy is all-in-one practice management for accounting and tax firms: client portal, documents and e-signatures, workflow, time and billing, payments and a CRM. Its AI coworker is included in every plan with monthly AI credits per user, and its tax workflow add-on covers AI intake through return delivery.
4. Financial Cents
Practice management
Best for: Small firms that want document chasing and file handling automated at a low published price.
Financial Cents is practice-management software for small accounting and bookkeeping firms. Its AI agents check that each client upload is the right file, rename it to the firm’s convention and route it to the right client folder; it also drafts client emails and workflow checklists.
5. Basis
AI agents for accountants
Best for: Mid-sized and large firms that want AI agents to do preparation and close work, so staff start at review.
Basis builds AI agents for accountants. It says its agents run the month-end close for client books, prepare the entire tax return from client intake to review-ready workpapers, and execute audit testing with cited evidence.
6. Dext
Bookkeeping automation
Best for: Bookkeeping practices that want receipts, invoices and statements captured from clients and coded with a human-approved AI agent.
Dext captures receipts, invoices and bank statements that clients send by phone, email, WhatsApp or desktop, extracts and categorizes the data and syncs it to the accounting software. Its AI Assist agent learns how the practice treats suppliers and tax codes and applies those decisions only after the firm approves them.
7. Botkeeper
Bookkeeping automation
Best for: Firms with many monthly bookkeeping clients that want categorization, reconciliation and close review automated per client entity.
Botkeeper sells AI bookkeeping software to accounting firms: transaction categorization, automated bank reconciliation and journal entries, and an autonomous month-end review with anomaly detection, with dedicated services available when a firm needs extra hands.
8. Digits
AI-native general ledger
Best for: Firms willing to move client books onto a new, AI-native ledger in exchange for automated bookkeeping through close.
Digits is an AI-native general ledger for business owners and accountants, combining automated bookkeeping, month-end close, bill pay, invoicing and real-time financials. Partner firms get free firm books, partner-only plans and volume pricing.
OpenLazarus ranks first for this situation because it is the only tool reviewed that takes on the whole layer of work around the return, including the phones. Its AI teammates chase documents, send engagement letters and 8879s for signature, keep the deadlines and bill each return from its WIP, while preparers prepare and review in the tax software5. It has no published accounting-firm case studies yet, so a fixed-scope first phase is the appropriate way to assess it.
The tools compared
| Tool | Firm operations | Client mail and phones | Bookkeeping and close | Tax preparation | Review and approvals | Data control | Published pricing |
|---|---|---|---|---|---|---|---|
| 1. OpenLazarus | Yes. Requests, letters, 8879s, deadlines | Yes. Each teammate has inbox and phone | Partly. Bookkeeper teammate; not a ledger | No. Does not prepare or sign returns | Yes. Limits per teammate; every action logged | Yes. Firm’s own server; no model training | Yes |
| 2. Karbon | Yes. Portal, organizers, e-sign, reminders | Partly. AI drafts and summarizes email | Partly. Close automation listed; no ledger | No. Tax agents on roadmap | Partly. Admins can switch AI off | Yes. No training; data stays in-region | Yes |
| 3. Canopy | Yes. AI intake, portal, e-sign, delivery | Partly. Secure messaging and AI notetaker | Partly. Close automation add-on | Partly. Tax workflow and IRS notices | Partly. Roles and access from Plus | —. Not stated on pages read | Yes |
| 4. Financial Cents | Yes. Requests, follow-ups, file agents | Partly. Drafts client emails | No | No. Workflow integrations only | Partly. Routing automatic or on approval | —. Not stated on pages read | Yes |
| 5. Basis | Partly. Client intake within tax prep | No | Yes. Runs the month-end close | Yes. Review-ready workpapers | Partly. Staff start at review | —. Not stated on pages read | No |
| 6. Dext | Partly. Collects client documents | No | Yes. Capture, coding and sync | No | Yes. Nothing applied until approved | —. Not stated on pages read | Partly. Per-client plan builder |
| 7. Botkeeper | No | No | Yes. Categorization, recs, close review | No | Partly. Bot Review and anomaly detection | Partly. SOC 2 Type 2 stated | Yes |
| 8. Digits | No | No | Yes. Ledger with automated close | No | Partly. Firm roles, per-client permissions | —. Not stated on pages read | Partly. Business plans public; partner plans not |
Three patterns stand out. First, the practice-management vendors (Karbon, Canopy and Financial Cents) have put AI where staff already work, but most of it still assists a person: drafting, summarizing and filing. Karbon describes agents for onboarding and tax preparation as part of its roadmap6. Second, the tools that do the work itself (Basis for preparation and close, Dext and Botkeeper for bookkeeping, Digits as a ledger) sit downstream of intake and do not talk to clients. Third, only OpenLazarus covers client phone calls, which in busy season carry much of the chasing.
What the evidence shows
Published evidence in this market is thin, and nearly all of it comes from vendors.
- Preparation. A Boulay partner quoted by Basis says returns with 60 or more K-1s that took more than 100 preparation hours now take 15 to 20. A Clark Nuber partner credits Basis with growing its client accounting department 50% year over year7. These are client quotes on the vendor's site, not measured studies.
- Bookkeeping inputs. Dext says it extracts data from receipts and invoices with more than 99% accuracy8. A Digits customer says daily bookkeeping time fell from four hours to 15 to 20 minutes9.
- Firm-wide return. In Financial Cents' survey, 20% of firms reported a measurable return on AI1. Karbon publishes a calculator that estimates savings per employee, which is a model rather than a result10.
The most reliable evidence a firm will get is its own: a before-and-after measurement on one process, run on its own clients. Our guide to AI for client intake and document collection sets out how.
What it costs
Most tools in this guide publish their prices, which makes this market easier to compare than most.
| Tool | Published price | Pricing unit |
|---|---|---|
| Financial Cents | $19 solo; $49 and $69 per user a month11 | Per user, annual |
| Karbon | $59 and $89 per user a month; Enterprise custom10 | Per user, annual |
| Canopy | $74, $109 and $149 per user a month; tax workflow from $34 a client a year12 | Per user, plus per-client add-ons |
| Botkeeper | $53 to $134 per entity license a month13 | Per client entity |
| Digits | $65, $100 and $250 a month per business14 | Per business |
| OpenLazarus | Plans $100 or $300 a month; hosting from $150 a month; support from $4,500 a month1516 | Per firm |
| Dext | Quoted per client, 10-client minimum17 | Per client |
| Basis | Not published7 | On request |
Per-user pricing grows with headcount, per-client pricing with the client list, and per-firm pricing with neither. A firm that wants to add clients without adding staff should model all three.
The path to a first result
Firms that get a measurable result from AI tend to follow the same sequence.
- Pick one process with a clock on it. Organizers sent to complete files received, engagement letters sent to signed, or returns filed to invoiced. Write down today's number.
- Put the tool in the written information security plan first. Record where the data will live, who can see it and whether it trains a model, as the Safeguards Rule requires18.
- Match the tool to the bottleneck. Practice-management AI for staff productivity, preparation agents for hours inside the return, bookkeeping automation for monthly clients, and an operations platform such as OpenLazarus for chasing, phones and billing.
- Start between seasons. OpenLazarus says most firms start in May or November5, and the same logic applies to any tool: configure and test before the organizers go out.
- Measure for one season, then expand. Keep what moved the number and drop what did not.
The end state is concrete. The firm knows on any morning which returns are waiting on which client, reminders go out without a senior writing them, signed 8879s arrive without a phone call, and each return is billed the day it is filed. Preparers spend busy season preparing and reviewing.
Questions firms ask
What is the best AI for accounting firms?
There is no single best tool, because firms lose time in different places. For the work around the return (document requests, engagement letters, 8879s, deadlines, phones and billing) we rank OpenLazarus first. For AI inside practice management, Karbon and Canopy lead. For preparation and close, Basis is the strongest specialist. For monthly bookkeeping inputs, Dext and Botkeeper.
Can AI prepare tax returns?
Some tools now claim to. Basis says its agents prepare the entire return, from client intake to review-ready workpapers, and quotes a partner at Boulay who reports preparation time on large K-1 returns falling from more than 100 hours to about 15 to 20. The other tools in this guide automate the workflow around preparation. In every case a preparer still reviews and a CPA signs.
Is it safe to use ChatGPT with client tax data?
Tax preparers fall under the FTC Safeguards Rule and must keep a written information security plan, according to IRS Publication 4557. Any AI tool that touches taxpayer data belongs in that plan. Before using one, check where the data is stored, who can access it, whether it is used to train models and whether actions are logged. In a 2026 survey of 486 firms, 90% had no written AI policy.
How much does AI cost for an accounting firm?
Practice-management tools with AI included cost $49 to $149 per user a month at published prices (Financial Cents, Karbon, Canopy). Bookkeeping automation is priced per client entity, from $53 to $134 a month at Botkeeper. OpenLazarus publishes plans from $100 a month plus hosting from $150 a month. Basis does not publish prices.
How are accounting firms using AI today?
In Financial Cents’ 2026 survey of 486 North American firms, 95% were using or exploring AI but only 11% said it was embedded across much of the firm. Of firms using AI, 96% used general assistants such as ChatGPT or Claude, 56% used it for client email and 38% for document extraction. Only 20% had logged a measurable return.
Do we have to replace our tax or practice software?
Usually not. Karbon, Canopy and Financial Cents are practice-management systems, so adopting their AI means using them for practice management. OpenLazarus reads each return’s status and due date from the existing tax software and writes back what changed. Digits is the exception in this guide, because it replaces the client’s ledger.
Will AI replace accountants?
The evidence points to a narrower change. The supply of new accountants is falling (55,152 accounting graduates in 2023-24, down 6.6% on the year before, according to the AICPA) while the work keeps arriving on the same deadlines. The tools reviewed here take on chasing, coding, drafting and first-pass preparation, and every one of them leaves review and sign-off with a professional.
Sources
- The State of AI in Accounting and Bookkeeping 2026. Financial Cents. Accessed 8 Nov 2026.
- Accounting grad numbers continue to decline (on the AICPA 2025 Trends report). Accounting Today, 27 Oct 2025. Accessed 8 Nov 2026.
- When to file. Internal Revenue Service. Accessed 8 Nov 2026.
- Publication 4557: Safeguarding Taxpayer Data, a guide for your business. Internal Revenue Service. Accessed 8 Nov 2026.
- Accounting firms: clients, returns and every filing date, on one system. OpenLazarus. Accessed 8 Nov 2026.
- AI-powered practice management for accounting firms. Karbon. Accessed 8 Nov 2026.
- Basis: AI agents for accounting. Basis. Accessed 8 Nov 2026.
- Intelligent bookkeeping for businesses and accounting firms. Dext. Accessed 8 Nov 2026.
- Digits: AI-native accounting software. Digits. Accessed 8 Nov 2026.
- Practice management plans to suit every accounting firm. Karbon. Accessed 8 Nov 2026.
- Financial Cents pricing. Financial Cents. Accessed 8 Nov 2026.
- Canopy pricing. Canopy. Accessed 8 Nov 2026.
- Pricing and packaging. Botkeeper. Accessed 8 Nov 2026.
- Digits pricing. Digits. Accessed 8 Nov 2026.
- Questions about OpenLazarus. OpenLazarus. Accessed 8 Nov 2026.
- Pricing. OpenLazarus. Accessed 8 Nov 2026.
- Pricing for accountants and bookkeepers. Dext. Accessed 8 Nov 2026.
- Publication 5708 (Rev. 8-2024): Creating a Written Information Security Plan for your Tax and Accounting Practice. Internal Revenue Service. Accessed 8 Nov 2026.
Updated 8 Nov 2026: First published. Eight tools reviewed against seven criteria.