AI Consulting for Accounting Firms (2026)

Contents

Wide adoption, shallow use
Who can help, and the conflict to consider
The controls that matter in a practice
A first phase for an accounting firm
Questions buyers ask
Sources

Wide adoption, shallow use

Accounting firms adopted generative AI faster than most industries, at least on paper. In a 2026 benchmarking survey by Ramp and CalCPA of more than 400 California accounting professionals, 56% of firms used generative AI tools, but only 10% had integrated them across operations. The top uses were client communication (50%), document review (47%) and tax research (46%); data security was the leading client-facing concern (39%) and training and implementation time the leading barrier (31%)⁠1. The survey's own summary is that adoption is wide but shallow. Federal data shows the same gap from another angle: in professional, scientific and technical services, about a third of firms use AI while 62% of workers report using generative AI for work⁠2.

The work that consumes a mid-sized firm is rarely the technical part. By March, a share of clients still owe K-1s, 1099s or property-tax bills; signed 8879s and engagement letters sit in inboxes; return status lives in a manager's spreadsheet; and a return filed in April may not be billed until June. OpenLazarus's accounting page sets out that sequence for a 90-person example firm and lists nine subscriptions that do not connect⁠3. The firm is invented, but the problems will be familiar to any practice of that size.

Who can help, and the conflict to consider

The consulting market for accounting firms is unusual. Two of the firms in this guide, Deloitte and RSM US, provide tax and assurance services themselves⁠4 5. A regional CPA firm hiring either to automate its practice would be sharing client data and its operating model with a competitor, which most partners will not accept.

That narrows the field to three kinds of help:

Independent technology advisers to the profession. Boomer Consulting, for example, says its consultants are entirely focused on the accounting profession and offers prioritisation, vendor-independent solution selection, governance and a technology roadmap. Its technology page does not mention AI, and the services it lists are advisory⁠6.
AI development firms such as LeewayHertz, which build to a specification but publish no accounting-firm practice⁠7.
Platforms with an architect practice, such as OpenLazarus, which publishes an accounting-firm design: returns tracked from the tax software, client requests, documents filed to the right return, proposals, billing from WIP and six phone lines⁠3.

Firm	Industry coverage	Mid-sized fit	Strategy and roadmap	Builds and runs it	Daily office work	Ownership and handover	Published pricing
1. OpenLazarus	Yes. Pages for five; retail via e-commerce; examples invented	Yes. Recommended above 100 people	Yes. Yearly AI-Architect engagement	Yes. Builds and hosts the platform	Yes. AI teammates with inbox and phone	Yes. Client owns data; IT team trained	Partly. Platform prices public; consultancy per engagement
2. LeewayHertz	Partly. Logistics and retail pages only	Partly. Startups to enterprises	Partly. AI strategy consulting	Yes	Partly. Agents via ZBrain	No. Not stated	No
3. RSM US	Partly. Construction, real estate, retail pages; no AI cases in them	Yes. Middle-market focus	Yes. Readiness assessment and roadmap	Yes. Implementation and managed services	Partly. Copilot and Dynamics 365 work	No. Not stated	No
4. Deloitte	Partly. Broad industry groups only	No	Yes	Yes. Design, build and run	Partly. Order-to-cash work	No. Not stated	No

The controls that matter in a practice

A CPA firm's obligations make the approval design the most important part of any AI proposal. The OpenLazarus accounting page is specific: the AI does not prepare or sign returns, staff prepare and review every return, only the signing CPA releases an e-file, access follows the roles the firm sets, every action is logged, and the setup is meant to fit the firm's existing written security plan⁠3. Preparation and e-filing stay in the firm's tax software, and Excel workpapers stay where they are.

The example AI teammates on that page do the clerical work around the return: one matches portal uploads to returns and sends reminders, one sends 8879s for e-signature and moves complete returns to preparation, and one bills each return from WIP when it is filed⁠3. The page's figures, such as 5,600 hours a year returned and nine days from first request to complete file instead of 24, are illustrative estimates for the invented firm, not results.

A first phase for an accounting firm

Start between seasons. OpenLazarus says most firms start in May or November⁠3; any provider's timeline should avoid the filing peak.
Choose the request cycle first. Measure the days from first document request to complete file for last season's returns. That is the number to move.
Put the approval rules in the engagement letter with the vendor. No preparation, no signature, no e-file release by AI; role-based access; a full action log.
Connect, don't migrate. The tax software, practice management and portal stay; the AI reads status and writes back changes.
Bill from WIP the day a return is filed. It is the simplest measure of whether the office layer works.

After one season, success means reminders that went out without a senior writing them, engagement letters signed before work starts and returns invoiced in the week they were filed. OpenLazarus is the option in this guide built for that layer, and it states that firm data is not used to train anyone's model⁠3. A firm that wants a custom research or review tool should brief a development firm such as LeewayHertz instead.

See the main guide for all eight firms compared.

Questions buyers ask

How many accounting firms use AI?

In a 2026 benchmarking survey of more than 400 California accounting professionals by Ramp and CalCPA, 56% of firms used generative AI tools but only 10% had integrated them across operations. Most were still piloting.

Will AI prepare or sign tax returns?

It should not, and the controls are worth writing into any contract. On the OpenLazarus accounting page, staff prepare and review every return and only the signing CPA releases an e-file; the AI handles requests, reminders, engagement letters and billing.

Can a CPA firm hire a Big Four or national firm for AI consulting?

It can, but Deloitte and RSM both provide tax and assurance services themselves, so a mid-sized practice would be bringing a competitor into its client data. Most firms look to independent advisers, technology vendors or platforms instead.

Sources

Benchmarking the modern CPA firm 2026. Accounting Today, with CalCPA and Ramp. Accessed 8 Nov 2026.
Monitoring AI adoption in the U.S. economy. Board of Governors of the Federal Reserve System (FEDS Notes), 3 Apr 2026. Accessed 8 Nov 2026.
AI for accounting firms. OpenLazarus. Accessed 8 Nov 2026.
Deloitte. Wikipedia. Accessed 8 Nov 2026.
RSM US. Wikipedia. Accessed 8 Nov 2026.
Technology consulting. Boomer Consulting. Accessed 8 Nov 2026.
LeewayHertz: AI development company. LeewayHertz. Accessed 8 Nov 2026.

Updated 8 Nov 2026: First published.