Our read
Start from the job. For audit engagements, test Fieldguide first: Aprio, Weaver and BerryDunn have published case studies, and Aprio built its own methodology into the agent. For tests of detail inside Excel, DataSnipper is listed by as many firms as Fieldguide, but only as OBSERVED logos, so ask for references. For screening a whole ledger for risk, MindBridge has KPMG as its one case. For tax research, start with Blue J, whose case studies include a firm of about 25 people. For recurring bookkeeping and close, Basis works on your current books at Boyum Barenscheer; Digits replaces the ledger and is for US entities only.
Which tool for which job?
| Your job | Where to start | The reference to read | What it documents |
|---|---|---|---|
| Run an audit or SOC engagement end to end |  Fieldguide |  BerryDunn | BerryDunn's SOC team handles client requests and reporting in Fieldguide, cut on-site travel by 75%, and now staffs a job with two remote people instead of four on site. 5 |
| Test documents and samples inside Excel |  DataSnipper |  JK Medora | A JK Medora partner says DataSnipper gave the team more time for critical issues and cash-flow suggestions to clients. The source names no specific audit procedure. 6 |
| Screen a full ledger for risk |  MindBridge |  KPMG | KPMG's global head of audit innovation says the MindBridge alliance brings more AI into the KPMG Clara audit platform for risk identification. 3 |
| Research a tax question with citations |  Blue J |  King & King | King & King, a Michigan firm of about 25 people, uses Blue J to check rulings and court cases and to take on technical work it used to refer out. 4 |
| Prepare recurring CAS journal entries |  Basis |  Boyum Barenscheer | Boyum Barenscheer has Basis read ADP payroll reports and prepare the journal entry for review: about four hours became 20 minutes. Accountants keep review and approval. 2 |
| Run client bookkeeping and the monthly close |  Digits |  Elev8 CFO | Elev8 CFO reports cutting its monthly close from 18 days to 5 on Digits, reclaiming nearly 90 hours a month across 27 client engagements without hiring. 7 |
What does a firm's name on a vendor page actually prove?
A named firm can mean two different things. A published case study or named quote (CONFIRMED) shows a task. A name on the vendor's customer list (OBSERVED) shows adoption and nothing more. Read every firm below with that split in mind. As of 2026-10-23, the usedby snapshot for accounting and audit in this category counts 107 companies: 24 CONFIRMED and 83 OBSERVED.
Within that sector, Fieldguide lists 32 firms (6 CONFIRMED), DataSnipper 32 (all OBSERVED), Blue J 23 (5 CONFIRMED), Basis 19 (2 CONFIRMED), Digits 10 (5 CONFIRMED) and MindBridge 5 (1 CONFIRMED). These are counts of listed firms, not market share or quality. DataSnipper's 32 are logos, so its count says little about which audit tasks firms run on it.
Which AI tool fits audit and assurance work?
Pick by where the work lives. Fieldguide is an engagement platform whose agents compile evidence and plans, test, check completeness and draft financial statements under practitioner oversight. Aprio, a firm of more than 3,400 people by its own case study, signed a proof of concept at the end of 2025 and asked Fieldguide to solve one real problem first: consolidations. Aprio's methodology is now built into Field Auditor, and auditors trusted it because the agent shows its reasoning. Aprio requires a team member to review and own anything an agent produces.
Weaver, working on governance, risk and compliance engagements, kept its proprietary methodology, configured Fieldguide around it and rolled out in phases. Weaver reports a 10-12% increase in realization in year one, and clients now log in to see progress instead of waiting for status calls. BerryDunn's SOC team is an older, workflow-centred case: 75% less on-site travel, two remote people instead of four on site, and 3-4 audits a week where one or two was the norm. BerryDunn also hired an Audit Coordinator to become the in-house Fieldguide expert, which freed manager time.
If your team works in Excel, DataSnipper is the closer fit. It runs inside Excel, matches and extracts data from source documents, and its Alwin agents run tests of detail, controls testing and disclosure checks with traceability for human review. DataSnipper says data is not used to train models by default and offers a data processing agreement. The proof is OBSERVED for every firm in the sector, plus a named quote from a JK Medora partner. DataSnipper's own customer page reports about 85% less testing time per selection in complex, high-volume scenarios; that is a vendor figure, not tied here to one firm.
MindBridge does a different job: it analyses the full population of transactions and prioritises risk with explainable results. KPMG's global head of audit innovation says the alliance brings more AI into KPMG Clara for risk identification. Cherry Bekaert, BDO, MNP and Citrin Cooperman appear as OBSERVED. Editorial recommendation: test it on a prior-year journal entry population you already understand, and compare what it flags with what your team found by hand.
Which AI tool should we use for tax research?
Start with Blue J if the job is answering tax questions with citations. It is a generative AI research tool whose answers rest on primary authorities, Tax Notes and IBFD content, covering US federal, state and local tax plus Canadian and UK tax. R2 Advisors, a boutique firm in McLean, Virginia, moved from CCH Answer Connect and Checkpoint because multi-layered entity work needed scenario analysis. It now runs several fact variations in one session with primary-law citations. With a single licence, R2 reports 6-8 hours saved per week for that user and 25% fewer non-billable research hours.
King & King is the small-firm example: about 25 employees in Imlay City, Michigan, which found Blue J through its state CPA association in early 2025. Its partner uses it to vet revenue rulings and court cases and to accept technical work, such as referrals from attorneys and financial planners, that the firm used to pass on. McKonly & Asbury and Crowe are also on the list, but only King & King and R2 Advisors are described here from their own case studies.
The data terms decide fit. Blue J stores and processes data only in the United States and offers no on-premise option. Uploaded files are deleted after 24 hours, and customer data is not used to train models by default. Its SOC 2 Type II report is available under an NDA. The Individual plan is listed at $1,498 per user per year, the Team plan is priced on request, and a free trial is offered. Coverage is US, Canadian and UK tax, so a firm outside those jurisdictions should check scope first.
Which tool for bookkeeping, client accounting and the monthly close?
Choose Basis if you want to keep your current ledgers and have agents prepare the work. Boyum Barenscheer, a Twin Cities CPA firm, started in February 2026 with six representative client books across different systems, then extended Basis to nearly all relevant CAS clients within six weeks. Vendor-reported results: cash and credit-card transaction coding fell from four hours to one, a payroll entry built from an ADP report from four hours to 20 minutes, and a monthly sales entry from one hour to 15 minutes. Accountants still own review, approval and client-specific decisions.
Basis also markets tax return preparation and audit testing, but the named case in these sources is CAS work at Boyum. On data, Basis lists SOC 2 Type II, ISO 27001 and ISO 42001, US-only hosting and no model training by default; its security page shows GDPR as coming soon.
Digits is a different decision because it is the general ledger itself, an AI-native one that categorizes transactions, reconciles accounts and automates close. Elev8 CFO reports a close cut from 18 days to 5, and Adept reports a 75% reduction in bookkeeping workload. Essentials is listed at $65 a month and Core at $100; Pro at $250 is shown as coming soon. A free trial exists. Digits states that its service is currently available only to US-based entities, which rules it out for many readers outside the US.
What are US accounting firms already using?
The US peers in the dossier, by tool and proof level: Steve Tonkin & Company (Digits, CONFIRMED), Schneider Downs (Basis, CONFIRMED), Anchin (Trullion, CONFIRMED), Countsy (Vic.ai, CONFIRMED), R2 Advisors (Blue J, CONFIRMED), Withum (Black Ore, OBSERVED) and BDO (DataSnipper, OBSERVED).
What this suggests: five of the seven US peers are CONFIRMED, so for Digits, Basis and Blue J you can ask the vendor for a reference call with a firm like yours. The two OBSERVED names, Withum on Black Ore and BDO on DataSnipper, show adoption only, so a reference call matters more there. Trullion, Vic.ai and Black Ore have no product source in this dossier and are not assessed here.
What transfers to a smaller firm?
Deployment shapes you can copy at small scale: Boyum's six-book pilot across different systems before widening; Aprio's single proof-of-concept problem on its own terms; R2 Advisors' single licence used on live client work; Weaver's phased rollout that kept its methodology; BerryDunn's named internal expert for the platform. This is our recommendation, not a vendor claim: pilot one job on one team, keep a named reviewer, and extend only after the output has been reviewed on real files.
What does not transfer: realization gains, close times, hours saved and travel cuts depend on each firm's starting process and are vendor- or customer-reported, not benchmarks. Aprio's co-build with a vendor and KPMG's alliance with MindBridge are enterprise arrangements. King & King shows that a firm of about 25 can use Blue J, but nothing here shows a firm that size running audit agents.
What the public record does not show
Prices for Fieldguide, DataSnipper, Basis and MindBridge are not published in these sources, so pricing is the first question for each vendor.
No figure here is an independent test: every saving is reported by the vendor or the customer. For DataSnipper, the sector entries are customer-list names without a described task.
Method and limits
2026-10-23
Scope: 107 companies classified in Accounting & Audit on usedby with a public line on a Finance & accounting AI tool. Every count on this page covers this sector only, not the tools' whole customer base. Company size is not recorded and is not claimed.
This dated snapshot counts distinct companies in public customer evidence. CONFIRMED means a case study, named testimonial or press source; DECLARED means the company itself told usedby; OBSERVED means a vendor listing; SIGNAL means a job post. All four count, but their strength differs. Removed and unlevelled lines are excluded. Examples are a capped selection verified within 90 days. Vendor facts are sourced records checked within 60 days. Coverage is uneven and missing evidence is unknown. These figures measure neither market share nor product performance; capture dates are not adoption dates.
Trullion
Vic.ai
Black Ore
