Caseware IDEA, Arbutus Analyzer and ACL Analytics are the three scripting-based, full-population testing tools that predate almost every audit-management platform in this guide, and they still get shortlisted together for exactly that reason: none plans an engagement, holds a signed-off workpaper or produces an audit-committee report, and all three exist to run a test against every record in a population rather than a sample. People search “IDEA vs ACL,” “caseware idea vs acl” and “arbutus vs acl,” never IDEA or Arbutus against Optro (formerly AuditBoard) or Workiva, because these three answer a different question than a GRC suite does. The complication a buyer needs going in: ACL Analytics is no longer really independent. Diligent owns it, prices it inside the Diligent One Platform, and the desktop application now authenticates only against diligentoneplatform.com, so a fair comparison has to weigh Arbutus and IDEA’s independence against ACL’s platform dependency, not just the three function libraries against each other.
This comparison covers the scripting engine and function library behind each, the one real ACL-to-Arbutus migration path and why IDEA has none, where SOX sits, which AI feature now traces to a named model, a side-by-side scorecard and fit-by-situation table, the thin public pricing evidence lined up together, and the Python and Power BI question: when a generalist data stack is the better buy than any of the three. It is part of the site’s independent buyer’s guide to internal audit software and follows the evidence rules in how we review audit software. Each tool also has its own full review: the Caseware IDEA review, the Arbutus Analyzer review and the ACL Analytics review. Not covered here: TeamMate Analytics and DataSnipper are Excel add-ins built to link evidence to a workpaper, not full-population scripting engines; the DataSnipper vs TeamMate Analytics vs Copilot comparison covers that decision separately.
Verdict
None of the three wins outright, and the real fork is commitment, not features. Choose ACL Analytics if you already run, or plan to run, the Diligent One Platform: AI Studio and Robots are genuinely native to it, and its AI layer is the only one now traceable to a named, dated model. Choose Arbutus Analyzer if you are an ACL shop wanting a lower-commitment alternative with no platform lock-in, but verify its ACL-conversion claims yourself, since the marketing language circulating about them is not what Arbutus’s own site states. Choose Caseware IDEA for the most affordable, ecosystem-independent seat with the deepest built-in function library and no interest in Diligent’s platform. A team with real Python or SQL skill and no budget for any of the three should read the Python and Power BI section below first.
Best for. Analytics-heavy audit functions of any size that need scripted, full-population testing rather than an audit-management platform; teams choosing among the three classic desktop scripting tools rather than a no-code AI layer alone; buyers who already know which ecosystem, Diligent’s or neither, they want to be inside.
Not for. Teams that want a SOX system of record, an audit-committee reporting layer or a case-managed issue tracker inside the analytics tool itself; buyers who need a clean, comparable per-seat price before a sales call, since none of the three reliably provides one.
Evidence. Research-based: vendor documentation and release notes, public procurement records, third-party pricing data, verified user reviews on Gartner Peer Insights and G2, and analyst coverage. We have not used any of the three products hands-on for this comparison.
Price evidence. None of the three publishes a price list. The only concrete public figures are a single UK National Highways license renewal for IDEA, £4,998.00 excluding VAT (Find a Tender Service, Notice 001028-2026, awarded 6 January 2026, seat count undisclosed); a third-party estimate for Arbutus from ITQlick of about $400 per user a year, not sourced to Arbutus; and, for ACL, only Diligent’s platform-wide Vendr marketplace median of $26,250 a year across 72 purchases, which is dominated by board-portal deals and should not be read as an ACL figure.
Last verified. 27 September 2026.
In this guide
- IDEA, Arbutus and ACL in one table
- Where they are different
- Head to head: the scorecard
- Fit by situation, side by side
- What each actually costs
- Migration between them
- The Python and Power BI question
- Our recommendation
- Questions about IDEA, Arbutus and ACL
- Sources and verification
- Related guides
IDEA, Arbutus and ACL in one table
All three show up on the same shortlist because all three do the same underlying job, script a test against a full data population rather than a sample, and none is an audit-management platform of the kind the site’s guide to the types of internal audit software describes. The table below is the fast version of everything that follows; figures and ratings come from the three individual reviews and the sources listed at the end of this page.
| Attribute | Caseware IDEA | Arbutus Analyzer | ACL Analytics (Diligent) |
|---|---|---|---|
| Owner | Caseware International; Hg Capital majority owner since late 2020 | Arbutus Software Inc., privately held, founder-led since 2003 | Diligent Corporation; backed by Insight Partners, Clearlake Capital and Blackstone |
| Headquarters | Toronto, Ontario | Burnaby, British Columbia (Metro Vancouver) | New York |
| Deployment | Windows desktop only; no browser or cloud-hosted analytics engine found | Desktop Analyzer, plus separately licensed HUB, WebConnect, Results Manager and Mainframe Server | Desktop Analytics app tied to the cloud Diligent One Platform; AI Studio and Robots are cloud |
| Core scripting engine | IDEAScript, plus Python through IDEA Lab | A proprietary 200-plus-function command language, plus Python for two machine-learning apps | ACLScript, plus a native AI command and, since version 19, a Python command |
| Built-in test library | 100-plus functions; SmartAnalyzer’s 90-plus tests; a 50-test Financial App | 200-plus functions; SmartApps prebuilt procedures across seven business-process domains (count disputed: nearly 70 versus nearly 90 on Arbutus’s own pages) | 100-plus claimed data connectors; scripted and AI-command testing, no published function count |
| ACL migration path | None stated anywhere in Caseware’s material | A named “ACL Projects” connector; no confirmed one-click or compatibility-percentage claim | Its own successor lineage; no import-from-a-rival path published |
| Named 2026 AI feature | AiDA, a paid add-on described only as in-app guidance; Caseware Verity is not shown to touch IDEA | Version 9’s AI and ML apps menu: OpenAI-based, opt-in, off by default; Python-based, offline | ACL AI Studio, cloud and no-code, now running on the Claude 4.6 model per Diligent’s release notes |
| Certifications found | ISO/IEC 27001, SOC 2 Type 2 and SOC 3, all scoped to Caseware Cloud, not the IDEA desktop application | None found anywhere, including the privacy policy, which offers only generic security language | ISO/IEC 27001:2013 and annual SOC 2 Type II at the Diligent One Platform level; GovCloud adds FedRAMP Moderate and DoD IL5 |
| Price evidence | A single UK license renewal, £4,998.00 excluding VAT a year, seats undisclosed | ITQlick’s unverified estimate of about $400 per user a year | No published figure; only Diligent’s board-portal-skewed Vendr median |
| User rating | TrustRadius 5.1/10 (10 reviews, IDEA-specific); G2 4.0/5 is suite-wide | G2 4.3/5 from only 2 reviews; no Gartner Peer Insights presence found | Folded into Diligent One Platform: G2 4.3/5 (154); Gartner Peer Insights 4.1/5 (148) in the Audit Management Solutions market |
Two things stand out before any feature-by-feature comparison. IDEA and Arbutus are the two genuinely independent products here, sold and licensed on their own; ACL is the only one whose desktop client is confirmed to require a specific cloud platform to function at all, and the only one of the three with a named, dated AI model behind it rather than general governance language. That trade, independence against a platform’s own AI investment, is the thread the rest of this comparison keeps returning to.
Where they are different
Where the three converge, we say so once and move on; what follows is where a buyer’s choice actually turns.
The testing engine: three scripting languages, one Python hedge
Each vendor’s claim to the category rests on its own scripting language and function library. IDEA pairs IDEAScript with more than 100 built-in functions, SmartAnalyzer’s more than 90 tests and a 50-test Financial App, read-only against an internal copy of the source file. Arbutus pairs its own command language, described on its site as the audit industry standard, with more than 200 functions and a SmartApps catalog of prebuilt procedures across seven business-process domains, though the exact count is disputed even on Arbutus’s own pages. ACL pairs ACLScript with a claimed 100-plus data connectors and, as of version 19, a native AI command for natural-language testing alongside the traditional script. All three also added a first-party Python surface within about a year of each other, a hedge the Python and Power BI section below returns to in full.
ACL compatibility: the one real migration path, mostly unverified
Arbutus’s whole pitch rests on being the ACL-compatible alternative, built around a named “ACL Projects” connection for converting a former ACL customer’s scripts. What could not be verified, including a targeted re-check of Arbutus’s own site, is the specific “one-click” framing or any compatibility percentage that circulates in reseller blogs, including a widely repeated but unlocatable 2021 quote attributed to founder Grant Brodie putting the figure at “at least 95 percent.” IDEA has no stated migration path from either rival anywhere in Caseware’s material, a real gap for a team consolidating from ACL or Arbutus, and ACL itself publishes no import path from a rival, unsurprising since it is the incumbent format the other two are chasing rather than the other way around. The full six-direction picture, including what a Python or Power BI move requires, is in the migration table further down this page.
SOX and controls: none of the three, by design
None of the three is marketed as a SOX or ICFR product. IDEA has no named SOX module; its one sourced controls-context example, the UK National Highways renewal, describes general “invoice gap detection and identification of control breakdowns,” not a labeled SOX engagement. Arbutus is the plainest of the three: no page uses the words “SOX,” “Sarbanes-Oxley” or “internal control framework,” and its Fraud and Compliance SmartApps cover sanctioned-party matching and keyword screening, not control walkthroughs. ACL AI Studio is the one exception: its July 2025 launch release names SOX explicitly as a use case and claims “AI Control Mapping and AI Control Generation capabilities,” but no control library or deficiency-classification workflow was confirmed inside ACL itself, so even that claim is a testing-layer claim, not a system-of-record one. Read all three as substantive-testing layers that can strengthen a SOX program run and evidenced elsewhere, never as the system of record for one; the site’s guides to SOX 404 and evaluating control deficiencies cover what that system of record actually needs to do.
AI: three stories, one now on a named model
| Aspect | Caseware IDEA | Arbutus Analyzer | ACL Analytics |
|---|---|---|---|
| Model or provider | Not named anywhere; general governance language only | OpenAI for the cloud apps; the machine-learning apps run offline with no provider to name | Amazon Bedrock hosts the models; ACL AI Studio specifically now runs on the Claude 4.6 model, per Diligent’s release notes for the product (17 September 2026) |
| On by default? | No; AiDA is a paid add-on, not bundled | No; the OpenAI-based apps are disabled and non-functional by default | Not stated; AI Studio is a separately licensed product |
| Named launches | AiDA, 28 May 2024, company-wide; Verity, 20 May 2026, not shown to touch IDEA | Version 9’s AI and ML apps menu, May 2025; non-English sentiment analysis added in version 9.1, December 2025 | The AI command, 4 December 2025; AI Studio generally available 25 September 2025; an Agentic AI mode toggle added 2 February 2026 |
| Data-use statement | General “governance, transparency and human oversight” language; nothing IDEA-specific found | The most specific of the three: a table’s data reaches OpenAI only once a feature is turned on; the machine-learning apps need no internet connection at all | By default, Diligent states it does not use AI models trained on customer data; AI-generated content is labeled |
IDEA’s AiDA is the thinnest of the three relative to its own company’s AI marketing: it is scaled down to “an in-app AI assistant for in-app guidance,” and Caseware’s flashier 2026 initiative, Verity, is not shown by any source read for this comparison to touch IDEA at all. Arbutus discloses the most about data handling of the three, splitting cleanly into cloud OpenAI apps that stay off until turned on and fully offline Python apps. ACL has the most real, dated motion on AI, and is now the only one of the three traceable to a specific foundation model with a specific date rather than a general oversight statement. Gartner’s April 2026 market guide for audit management software warns buyers to be wary of agent-washing; measured against that standard, ACL’s AI Studio and command are the most concretely documented of the three, though the sound response to any of them is still to ask for the labeled output and audit trail on one real test before taking a demo’s claims as settled.
Reporting, dashboards and the audit-committee gap
IDEA produces interactive dashboards, a PDF Report Reader, and export to Power BI, Tableau or Excel; there is no audit-committee report generator. Arbutus’s version 8 added visual Compare, Classify, Stratify and Aging outputs and drill-down pivot tables, WebConnect adds browser access for non-technical stakeholders, and Results Manager supplies its own dashboard; again, no committee report generator was found. ACL has no reporting layer of its own: the Reports app it inherited from HighBond reaches end of life 30 September 2026, replaced by Activity Center, and the September 2026 conversational AI reporting release is a Diligent One Platform feature, not an ACL-specific one. None of the three should be bought expecting the board pack an audit-native platform ships; hold whatever a demo shows against the site’s audit committee deck template.
Security, certifications and vendor viability
Security coverage is uneven, and in every case (see the at-a-glance table above) it sits above the product rather than on it directly: IDEA’s certifications belong to Caseware Cloud, a different product; Arbutus has no certification claim of any kind, including in its privacy policy; ACL’s belong to the Diligent One Platform as a whole, where GovCloud adds FedRAMP Moderate and DoD IL5 without clearly listing AI Studio. ACL is the best-documented of the three on paper for exactly that reason, not because Diligent has certified ACL Analytics itself as a standalone product. Vendor viability for all three, Hg Capital’s 2025 C-suite turnover at Caseware, Arbutus’s thin press trail, and Diligent’s explored-but-uncompleted 2024 sale, is in the scorecard’s own row further down. A security review for any of the three should ask the vendor to state, in writing, which certification actually covers the software you are buying.
Implementation, training and the administrator question
None of the three publishes a real implementation timeline. IDEA bundles a Passport user portal, tutorials, videos and sample scripts, plus instructor-led training with no published hours or service-level agreement. Arbutus offers live virtual training through Arbutus Certified Trainers carrying NASBA continuing-education credit, and claims, in a vendor self-comparison we could not independently verify, that its own implementation costs less than “alternative solutions.” ACL publishes no training curriculum either; what is documented instead is a forced-upgrade cadence to staff for: support for version 16.x ended 3 February 2025, support for 17.x ended the same day version 19.0 shipped, 4 December 2025, and any on-premise Robots Agent must upgrade in lockstep with Analytics or lose encryption compatibility. Whichever of the three you pick, budget an administrator’s time as a real, recurring cost, not a one-time setup fee.
Head to head: the scorecard
The 12 areas and four levels are the ones used across every review and comparison in this guide, so the columns below can be read against any platform elsewhere in the series. All three are analytics tools rather than audit-management platforms, so several areas are marked not applicable rather than forced into a rating none of the three was built to earn.
| Area | Caseware IDEA | Arbutus Analyzer | ACL Analytics |
|---|---|---|---|
| 1. Risk assessment and planning | n/a | n/a | n/a |
| 2. Engagement workflow | n/a | n/a | n/a |
| 3. Workpapers and evidence | Limited | Limited | Limited |
| 4. Issues and follow-up | n/a | Adequate | n/a |
| 5. Reporting | Adequate | Limited | n/a |
| 6. SOX and controls testing | Limited | Not offered | Limited |
| 7. Analytics and automation | Strong | Strong | Strong |
| 8. AI features | Limited | Adequate | Adequate |
| 9. Quality program support | n/a | n/a | n/a |
| 10. Auditee experience | n/a | n/a | n/a |
| 11. Administration, integrations and security | Limited | Limited | Adequate |
| 12. Cost and contract | Limited | Limited | Limited |
| Vendor viability | Adequate | Stable but opaque | Moderate roadmap risk |
Two things are worth calling out rather than passing over silently. First, area 2, engagement workflow, reads not applicable for all three: none of them ships a planning memo, a risk and control matrix or a review sign-off object of its own, and Arbutus’s Results Manager and ACL’s AI-recommended tests come closest without actually amounting to a built and signed-off engagement. Second, analytics and automation, area 7, is the clearest positive tie in the table: all three earn Strong there on real, differentiated capability, rather than the shared absence behind areas 1, 2, 9 and 10. Buy any of the three for full-population testing, and expect to look elsewhere entirely for rows 1, 2, 4, 5, 9 and 10 as they apply.
Fit by situation, side by side
The eight situations are the same on every review and comparison in this guide.
| Situation | Caseware IDEA | Arbutus Analyzer | ACL Analytics |
|---|---|---|---|
| First system for a small team (1 to 5 auditors) | Strong fit | Strong fit | Workable |
| Mid-size function (6 to 25 auditors) | Strong fit | Strong fit | Strong fit |
| Large or global function (25+ auditors) | Strong fit | Strong fit | Strong fit |
| SOX-heavy public company | Workable | Workable | Workable |
| Bank or credit union | Strong fit | Strong fit | Strong fit |
| Public sector, higher education or nonprofit | Strong fit | Workable | Strong fit |
| Analytics-heavy team | Strong fit | Strong fit | Strong fit |
| Consolidating GRC across the three lines | n/a | n/a | n/a |
Only two rows separate the three. ACL alone rates Workable rather than Strong for a small team’s first system, the fair price of quote-only pricing with no small-team edition and a product that rewards ACLScript fluency a very small team may lack; IDEA and Arbutus both rate Strong there instead, mainly on the strength of one relatively inexpensive seat. Arbutus alone rates Workable rather than Strong for public sector, higher education or nonprofit work, since it has no SOC 2, ISO 27001 or FedRAMP claim for a security review to point to, where IDEA’s public-sector fit rests on a single UK government renewal and ACL’s rests on a federal case study plus GovCloud’s platform-level FedRAMP Moderate and DoD IL5 authorizations. Every other row ties, and the SOX-heavy row ties at Workable for all three, the table’s own confirmation that none belongs at the center of a 404 program.
What each actually costs
A real five-year total cost of ownership needs a clean per-seat price to multiply, and none of the three gives you one. What follows uses only the public figures already cited, with the assumptions it takes to get any dollar figure at all labeled plainly; treat every illustrative number as a floor built on a guess, not a quote, and get your own written price before budgeting from any of it. The site’s internal audit software pricing guide has the negotiation script that applies across all three.
| Line | Caseware IDEA | Arbutus Analyzer | ACL Analytics |
|---|---|---|---|
| Best public figure | £4,998.00 excluding VAT a year (National Highways renewal) | About $400 per user a year (ITQlick’s own unverified estimate) | No product-specific figure; Diligent’s platform-wide Vendr median is $26,250 a year |
| What it actually covers | One organization’s license, term 1 February 2026 to 31 January 2027; seat count not stated | Analyzer only; HUB, WebConnect, Results Manager and Mainframe Server are separate, additional licenses | 72 analyzed Diligent deals, dominated by board-portal customers, not audit or analytics buyers specifically |
| Illustrative annual figure (labeled assumption) | Not computable per seat; treat the renewal as one undisclosed-seat-count total, not a per-seat rate | $2,000 a year, assuming 5 users at ITQlick’s estimate, Analyzer only | Not computable from public data; do not scale the Diligent median into an ACL estimate |
| Five-year total, flat, no increase modeled | £24,990.00 excluding VAT, on the same undisclosed-seat-count basis | $10,000, on the 5-user, Analyzer-only assumption above | Not computable from public data |
| Known cost not included above | The AiDA add-on fee; Passport training; support beyond what the brochure describes | HUB, WebConnect, Results Manager and Mainframe Server, each licensed and quoted separately | AI Studio as a possibly separate line from Analytics; a Diligent One Platform subscription, if the desktop tool cannot be bought without one; Robots capacity |
Three cautions apply equally. The IDEA figure is a single overseas renewal with an undisclosed seat count, not a rate card; multiplying it by five shows that one buyer’s total commitment, not what a seat costs you. The Arbutus figure rests entirely on a third-party estimate ITQlick itself never had Arbutus confirm, multiplied by an assumed head count we chose for illustration. And the ACL row is deliberately left blank rather than filled with Diligent’s board-portal-weighted median, since its own review says plainly that is not an ACL benchmark; a blank cell is more honest than a number that would look precise and is not. Get an itemized, written quote from all three before you compare anything past this table.
Migration between them
Only one of the six possible directions between these three products has a named, vendor-built connector; the rest are manual rebuilds.
| Migration path | What exists | What is unconfirmed or missing |
|---|---|---|
| ACL scripts or projects into Arbutus | A named “ACL Projects” connection, built specifically for this move | No “one-click” claim or stated compatibility percentage on Arbutus’s own site; a widely repeated 2021 “at least 95 percent” quote could not be independently located |
| ACL scripts or projects into IDEA | Nothing found anywhere in Caseware’s material | Plan to rebuild test logic in IDEAScript or Python rather than convert it |
| IDEA or Arbutus logic into ACL Analytics | Nothing published; ACL’s own recent history is about migrating away from its own legacy pieces, not importing rivals’ | The AX server, Direct Link and the bundled Access Database Engine are things ACL customers have had to migrate away from, not evidence of an import path |
| Any of the three into Python or Power BI | Ordinary exports, CSV, Excel and ODBC, are the practical bridge all three support | None publishes a structured project-export format a Python or Power BI pipeline reads directly; test logic gets rebuilt as code, not converted |
The asymmetry is worth remembering on its own: an ACL shop shopping for an alternative has one real, if unverified-in-detail, path, into Arbutus. A team on IDEA that later wants to switch to either rival, or a team on any of the three that wants to move into a generalist stack instead, should budget the move as a rebuild of test logic, not a conversion of it, the same caution the site’s implementing audit management software guide gives for platform migrations generally.
The Python and Power BI question
None of the three packaged tools has a moat a competent Python or SQL analyst could not eventually replicate with pandas, a scheduled notebook or a warehouse query, and the vendors’ own Python hedges above suggest they know it. That is a signal about where the real skill in this market sits, not a reason to skip a packaged tool by default.
- A generalist stack tends to win when the team already has real Python or SQL skill and no budget for a per-seat license; when the organization is already centralizing data in a warehouse such as Snowflake or Databricks that any of the three would only connect to anyway; or when the function is small enough that avoiding a license fee outweighs maintaining code someone has to keep alive.
- A packaged tool tends to win when a non-coding auditor needs to run a test directly, through SmartApps, IDEA’s library or AI Studio’s no-code layer, without waiting on a scripter; when the read-only source protection and immutable run history IDEA and ACL build in by default matter to an examiner, since a hand-rolled notebook has to earn that trust from scratch; or when a built-in function library, Benford’s Law, gap detection, fuzzy duplicate matching, saves real time against writing the same tests by hand.
Cost is not a wash between the two paths. Python itself carries no license fee, only analyst time; Microsoft’s Power BI Pro is $14 a user a month and Premium Per User is $24, with Copilot inside Power BI needing Fabric F2 capacity, and Microsoft 365 Copilot runs $30 a user a month on top of that. Alteryx, the closer packaged competitor to a coded pipeline, prices its entry tier at $250 a user a month for one to 10 users, cloud-only, excluding Ask Alteryx, Auto Insights and Orchestrator. Set against IDEA’s undisclosed-seat renewal or Arbutus’s unverified $400-a-year estimate, a generalist stack built on Power BI alone can be cheaper per seat than any of the three, before counting the analyst time a built-in library exists to save. The site’s Alteryx vs Power Query and Power BI comparison and the audit analytics software compared guide work through that build-versus-buy decision further, and the site’s guide to RPA in internal audit covers the scheduling layer a homemade pipeline still needs to replicate what HUB or Robots do out of the box.
Our recommendation
Start from your platform commitment, not a feature checklist, since the scorecard above shows the three tying on far more rows than they differ. Already running, or buying into, the Diligent One Platform: read the ACL Analytics review first, since AI Studio and Robots are native to that ecosystem and its AI layer is the only one now traceable to a named model, but get the platform-subscription dependency and the AI Studio GovCloud question answered in writing before you sign.
Already an ACL shop wanting a lower-commitment alternative, or simply the closest thing to ACL without Diligent’s platform attached: start with the Arbutus Analyzer review, and insist on converting a real project of your own in the demo rather than taking the ACL-compatibility marketing at its word.
Neither ecosystem matters to you, and you want the most affordable, independent seat with the deepest built-in function library: read the Caseware IDEA review; budget real training time for the advanced end of the library, and do not expect Caseware’s Verity AI story to reach the product yet.
None of the three replaces an audit-management platform. A function that also needs a SOX system of record, a committee reporting layer or a case-managed issue tracker should look at Optro, Workiva or the wider Diligent One Platform instead, and treat whichever analytics tool it buys as the layer underneath one of those; the site’s best internal audit software ranking places all of these against the full field.
Questions about IDEA, Arbutus and ACL
IDEA vs ACL: which is better for a team that already has ACL scripts?
Neither, cleanly. IDEA has no stated path for importing ACL scripts, so a team migrating off ACL should look at Arbutus first, whose named ACL Projects connector is the one real, if not independently verified for its exact compatibility, migration path among the three. Staying on ACL Analytics is the only option needing no conversion, at the cost of continued Diligent One Platform dependency.
Is Arbutus’s one-click ACL conversion claim real?
Arbutus supports a named ACL Projects connection, so converting logic from ACL is a genuinely supported path, not a myth. What could not be verified, including on Arbutus’s own site, is the specific “one-click” framing or any compatibility percentage, including a widely repeated but unlocatable 2021 “at least 95 percent” quote. Convert a real project of your own in a demo before relying on either claim.
Which of the three has a real, dated AI feature, not just marketing language?
ACL AI Studio: generally available since 25 September 2025, an agentic mode toggle added 2 February 2026, and, per Diligent’s own release notes, running on the Claude 4.6 model since 17 September 2026. Arbutus is the most transparent about data handling, splitting into opt-in cloud OpenAI apps and fully offline Python apps. IDEA’s AiDA is the thinnest of the three: a paid add-on described only as “in-app guidance,” with Caseware’s larger 2026 AI initiative, Verity, not shown by any source to touch IDEA.
Which of the three is cheapest?
There is no clean answer, because none publishes a real price. IDEA’s only figure is a single UK government renewal with the seat count undisclosed; Arbutus’s is an unverified third-party estimate of about $400 per user a year; ACL has no product-specific figure at all, only Diligent’s board-portal-dominated Vendr median. Get a written, itemized quote from all three before assuming any of them is the affordable option.
Do any of the three replace a SOX or audit-management system?
No. All three are analytics engines that test data, not systems that hold a control library, a signed-off workpaper or an audit-committee report. ACL AI Studio’s launch release is the only one to name SOX explicitly as a use case, but even that is about AI-recommended tests and control mapping, not a system of record. Pair whichever you choose with a platform such as Optro, TeamMate or the wider Diligent One Platform for that layer.
When should we use Python or Power BI instead of any of these three?
When the team already has the scripting skill, has no budget for a per-seat license, or is already centralizing data in a warehouse the packaged tools would only connect to anyway. What you give up is the read-only source protection and immutable run history IDEA and ACL build in by default, and the no-code path that lets a non-coding auditor use SmartApps, IDEA’s built-in library or AI Studio directly. The Python and Power BI section above works through the trade-off in full.
internalauditguide.com has no commercial relationship with Caseware, Arbutus Software, Diligent, or any other vendor named on this page. We take no vendor money, run no affiliate links and accept no sponsored placements, and no vendor saw this page before publication. Product and company names are the trademarks of their owners. Corrections: desk@internalauditguide.com.
Sources and verification
- Caseware, IDEA product page — five-step workflow, function counts, AiDA-as-add-on language (accessed 27 September 2026).
- UK Find a Tender Service, Notice 001028-2026 — the National Highways IDEA renewal figure, term and award date (accessed 27 September 2026).
- Caseware, Verity launch release — Verity’s scope and the absence of any IDEA mention (accessed 27 September 2026).
- Arbutus Software, Arbutus Analyzer — function count, SmartApps count, the ACL Projects connector, read-only access (accessed 27 September 2026).
- Arbutus Software, About Us — founder and CEO Grant Brodie, company background (accessed 27 September 2026).
- Arbutus Software, AI/ML FAQs — the OpenAI-based versus offline-Python split and the data-use statement (accessed 27 September 2026).
- ITQlick, Arbutus — the “about $400 per user a year” third-party estimate (accessed 27 September 2026).
- Diligent, ACL Analytics 19.0 release notes — version 19 features, the domain restriction and the encryption change (accessed 27 September 2026).
- Diligent, ACL AI Studio release notes — the Agentic AI mode toggle (2 February 2026) and the Claude 4.6 model statement (17 September 2026) (accessed 27 September 2026).
- Diligent, ACL AI Studio launch release — the 15 July 2025 announcement, features and quote (accessed 27 September 2026).
- Diligent, trust and security page — certifications, data residency and the AI data-use and Amazon Bedrock statements (accessed 27 September 2026).
- Diligent, pricing page — the absence of published pricing and the request-a-quote model (accessed 27 September 2026).
- Vendr, Diligent marketplace page — the median and range figures and the board-portal caveat (accessed 27 September 2026).
- G2, Diligent One Platform reviews — the rating, review count and the themes naming ACL Analytics (accessed 27 September 2026).
- Gartner Peer Insights, Audit Management Solutions market — the 4.1-from-148 rating used for ACL, and the confirmed absence of a standalone IDEA or Arbutus listing (accessed 27 September 2026).
Related guides
- Internal audit software: the independent buyer’s guide — every review, comparison and buying guide in one place.
- How we review audit software — the evidence levels, the scorecard and the fit-by-situation method.
- The audit software shortlist finder — eight questions, a shortlist with the reasons from each review.
- The requirements matrix — 156 weighted requirements and vendor scoring in a free Excel workbook.
- Caseware IDEA review — still the default audit analytics tool?
- Arbutus Analyzer review — the ACL-compatible analytics tool, explained.
- ACL Analytics review — version 19, AI Studio, Robots and what replaced AX.
- Diligent One Platform review — HighBond, ACL and the platform ACL Analytics now lives inside.
- Best internal audit software — 25 platforms and tools compared by use case.
- Types of internal audit software — where audit analytics sits among audit management, GRC, SOX and compliance tools.
- Internal audit software pricing — real numbers, pricing models and how to negotiate.
- Selecting an audit management system — the vendor-neutral RFP method.
- The audit software demo script — 25 scenarios, including the ACL-conversion test worth asking for.
- 15 mistakes internal audit teams make when buying software — and how to avoid each.
- Evaluating AI in audit software — what is real and what is agent-washing.
- Alteryx vs Power Query and Power BI — which data stack an audit team should buy instead.
- Audit analytics software compared — IDEA, Arbutus, ACL, Alteryx, Power BI and Python side by side.
- DataSnipper vs TeamMate Analytics vs Copilot — the adjacent decision, Excel evidence add-ins rather than scripting tools.
Leave a Reply