Audits per year
The plan covers what the team can staff. The rest of the audit universe waits for next year's cycle, and often the one after.
For internal audit
Atorian is the AI operating system for internal audit. It does the audit work itself: it interviews auditees, reads every document, tests full populations of data, and drafts findings that trace back to their evidence. Your team covers more of the audit universe, more often, without growing in proportion.
Control AC-04Leaver accounts are disabled within one business day
Enough to fix it: every account, by name.
01The constraint
How many audits fit in a year. How many transactions a test can sample. How long fieldwork can run. The traditional audit model was built on the limits of human labor, which makes a risk-based plan, in practice, a capacity-based one.
Audits per year
The plan covers what the team can staff. The rest of the audit universe waits for next year's cycle, and often the one after.
Items per test
A sample of 25 stands in for a population of thousands. It can tell you something is wrong. It can't tell you where.
Hours per audit
Fieldwork ends when the budget does, not when the questions do. The areas that deserved a second look rarely get one.
02What's different
Audit management software now uses AI to draft the plan, tag issues and chase follow-ups. That makes the paperwork faster. The fieldwork itself (the interviews, the evidence, the testing and the judgment about what it all means) is still done by hand. That is the part Atorian does.
AI around the audit
Atorian, inside the audit
Keep your audit management system. Atorian does the fieldwork it was never built to do.
03How it works
Atorian follows an engagement the way your team runs one. Your auditors set the scope, review the work and own every conclusion. The platform does the legwork, at a scale no team can staff.
Atorian builds a profile of the audited unit from its own documents and public sources, and names the source of every fact. It proposes the fieldwork plan and the standards that apply, and shows why each one does.
Your teamapproves the scope and the standardsProcess owners answer in a guided workspace instead of a long questionnaire. They can talk instead of typing, or hand over a document and let it answer the questions it covers. Atorian asks follow-up questions based on what they say.
Your teamchooses who is interviewed, and reads every answerPolicies, procedures, committee minutes, prior reports, system exports and logs. Every document is read in full, every quote is checked against its source text, and every claim records who made it.
Your teamadds anything the auditee didn't sendAudit apps run full-population tests on the data you already export: every leaver against active accounts, every loan approval against credit policy, every patch against its deadline. You get the exceptions, not ten thousand rows.
Your teamdefines the test and the thresholdObservations become candidate findings, each traced to the quotes and documents it rests on. Where an interview and the evidence disagree, Atorian flags it, and the finding can't be confirmed until the contradiction is resolved. Every observation is either part of a finding or listed with the reason it isn't.
Your teamconfirms, challenges or rejects each findingA draft report is ready before the full analysis has run. Recommendations carry timeframes, and every finding opens to its evidence, so the report holds up when management pushes back.
Your teamsigns the report"Which of these findings rests on a single interview?"
The AI Advisor reads the whole engagement, interviews and attachments included, and answers questions about it while the work is still underway.
04Evidence
Select a source to see what it says and which part of the finding it supports.
Illustrative finding. The organization, figures and quotes are fictional, written to show the format.
05The audits
Atorian is an operating system for audit. We started with technology audits, where the evidence is densest and the specialists are hardest to hire. The library keeps growing, and any audit in your plan can be built on it with us.
Bring an audit from your plan, whether it covers procurement, credit, operations or anything else, and we build it into the platform with you, against your own policies and methodology.
Some tests are too large to do by hand. An audit app automates one of them on the data you already export, and returns the exceptions straight into the audit.
5,000,000
login records, with 100,000 exceptions in them
One line in the report
2% of logins without MFA. 7% from risky devices, none of them blocked.
Every leaver, against every account that is still active.
HR leavers file × directory export
Every approval, checked against your own credit policy.
Loan approvals × credit policy
Every critical vulnerability, against the deadline it was given.
Scan exports × patching SLAs
Every supplier you pay, against the third-party risk register.
Vendor master × risk register
06The outcome
Whole populations instead of samples, so a finding names every exception rather than estimating them.
Test key controls more often, and move from point-in-time audit toward continuous assurance.
Go further wherever the evidence points to risk, without reopening the budget to do it.
Cover more of the audit universe without growing the team in proportion.
The business and technology teams answer in their own time and hand over documents instead of sitting through meetings.
Management and the audit committee get a current, evidence-based view of risk, not last year's.
Auditors working alongside AI agents, covering far more of the organization's risk, and spending their time on judgment, complexity, and the questions that truly need human expertise. That is the function we are building for.
Menny Barzilay, co-founder and CEO, former chief IT auditor07Security
You'll be trusting the platform with policies, evidence and interview records. Your CISO will want to review it, and should. Here is what they'll find.
Isolation
Each customer runs in its own instance, with its own database, its own storage and its own users.
Deployment
A dedicated cloud instance, or an installation on premises in your own environment, with AI calls routed through your own gateway.
Traceability
Every finding traces to the quotes it rests on, and every quote to its source document and to the person who said it.
Testing
An independent penetration test in July 2026 found no critical or high-severity issues. SOC 2 Type I, with Type II in progress.
Access
Two-factor authentication, single sign-on with Google, and protection against brute-force attempts.
Before approval
Security approval usually takes longer than the decision. The first stage can run in Atorian's own environment, so your team sees the platform before your process has to approve anything.
08Getting started
A fixed-scope pilot on a real audit, with your own data and your own auditors. Scope, sources, deliverables and success measures are agreed before it starts.
Route A
Pick an audit from this year's plan. Atorian runs it from planning to report, with your auditors working in the platform alongside it.
Route B
Pick a single use case, usually a full-population test you have only ever sampled, and we build the audit app for it.
09The team
Co-founder and CEO
Former chief IT auditor and former CISO. Has chaired the audit committees of two nonprofits, and lectures in IIA Israel's auditor training. CTO of the Blavatnik Interdisciplinary Cyber Research Center at Tel Aviv University.
Chief Services Officer
Former partner at PwC and KPMG. Founded a company that was acquired by Mastercard.
Co-founder and CTO
Fifteen years at Check Point, building some of its most successful products.
Advised by Mark McLaughlin, chairman of the board at Qualcomm, Steve Zalewski, former CISO at Levi's, and Prof. Isaac Ben Israel, director of the Blavatnik Interdisciplinary Cyber Research Center.
Book a walkthrough
Thirty minutes with the people who built the platform. Walkthroughs are run by practitioners, not a sales script, so bring your hardest questions.