The rules that bind you
Binding law and supervisory expectation. Where one of these applies to your system, the audit is scoped against its named articles - not against a summary of them.
- AI regulation 9
- Model risk - banking supervisors 3
Find the AI risks no one’s checked yet.
iDharma is an independent AI auditor. We test AI that decides about people, and AI agents that act on their behalf, against the rules that apply.
Independently verified
No payment until scoped
1–4 week reports
Every finding mapped to a named framework
A named expert on every audit
No payment until the scope is agreed
Prioritised report in 1-4 weeks
No black box. No vendor spin.
Our method
Every finding points back to evidence
We agree what is being examined before anything starts, test it against named standards rather than an in-house checklist, and tie every conclusion to something you can inspect.
That is what separates an audit from an opinion. What you get is a document you can put in front of a board, a customer or a regulator - and show your working for every line in it.
Five dimensions
The five ways your AI gets judged
AI systems are only as trustworthy as the questions you ask. We evaluate performance, behaviour and governance across five critical dimensions.
-
Accuracy
Does it actually work?
Real outputs tested against ground truth, measuring how often it gets the answer right and where it fails.
Measured -
Fairness
Does it treat everyone the same?
Bias probed across every group it touches, with the outcomes for each group compared side by side.
Impact ratios -
Security
Can it be manipulated?
Attacked the way a bad actor would, through prompt injection, jailbreaks and manipulated inputs.
Adversarial -
Privacy
Is personal data protected?
What it stores, shares, or is tricked into revealing, traced from the data it takes in to the answers it gives.
Data flows -
Compliance
Would it pass a regulator’s review?
Checked against the rules that actually apply to you, with every gap mapped to a named framework.
29 frameworks
Every gap mapped to
NIST AI RMF
ISO 42001
EU AI Act
HIPAA
India DPDP
SOC 2
Free self-check
Not sure where your AI stands?
Five quick questions, one real finding about your AI - free, in 60 seconds.
-
Answer five questions
60 seconds - no signup, no sales call, no pitch
-
Get a straight read
Across governance, data, bias, security and compliance
-
Act on it
A specific, real finding about your AI - yours to keep
AI Risk Snapshot
Sample
- Governance Medium
- Data provenance High
- Bias & fairness Low
- Security Medium
- Compliance High
Overall exposure
High
Request your audit
Answer 5 questions - get a real, specific finding. 60 seconds. No signup.
Domains we audit
Built for how AI actually fails, by industry
Domain 01
Fintech
Lending, credit scoring, and fraud detection, audited for fair-lending exposure and model risk.
Request an auditDomain 02
Finance
Trading, portfolio, and financial-reporting AI, reviewed for investment and reporting risk.
Request an auditDomain 03
Healthcare
Clinical and patient-facing AI, reviewed for PHI handling and emerging state AI rules.
Request an auditDomain 04
Insurance
Underwriting and claims automation, tested for proxy discrimination and fairness.
Request an auditWhat every engagement includes
The four things every audit gives you
The method, the independence, the report, and the terms - set out before anything is charged.
-
The Standards
Findings trace to six named, publicly recognised frameworks rather than an internal checklist of our own.
NIST AI RMF ISO/IEC 42001 EU AI Act HIPAA SOC 2 India DPDP -
The Independence
We audit AI we did not build. No vendor relationship, no conflict of interest, and evidence-based findings instead of a self-graded checklist.
-
The Report
Clear findings with a prioritised fix list, typically one to four weeks from scope to report. It reads plainly instead of burying you in a data dump.
-
The Terms
Nothing is charged until you approve the audit scope. You agree what the work covers before anything is billed.
What we do
AI audits, explained plainly
No jargon and no sales pitch. Just what an AI audit is, why it matters, who needs one, and how iDharma does it.
An independent review of the AI systems you already use, checking how accurate, fair, secure, and compliant they really are rather than taking the vendor’s word for it.
Unaudited AI is unmeasured risk. Bias, security gaps, and compliance failures stay invisible until they cost you in fines, lost trust, or a regulator asking questions you can’t answer.
Any business running AI in a regulated or customer-facing process: lending, healthcare, hiring, or customer service, especially under EU AI Act, HIPAA, or SOC 2 obligations.
We independently test your AI against recognized standards like NIST AI RMF, ISO 42001, and the EU AI Act, then hand you a clear, prioritized report. Named auditor, cited methodology, no black box.
Why iDharma
An independent read you can stand behind
Independent by structure
A verified expert who didn’t build your AI is the one who reviews it. Independence isn’t a claim we make about ourselves - it’s the arrangement that makes a finding defensible to a board or a regulator.
Standards, not say-so
Every finding traces to a named framework - NIST AI RMF, ISO/IEC 42001, the EU AI Act - never an internal checklist of ours. Soundness you can point at is part of being trustworthy, not a separate exercise.
Clarity over cleverness
A report you have to decode has failed. Risks, compliance gaps and a prioritised fix list, in plain language, with the thing that matters most obviously the thing that matters most.
Built in, not stickered on
Trust here is shown, not decorated: no hollow badges, no invented scores, no countdown pressure. See the work before you pay - the free 60-second Risk Snapshot hands you a real finding, no commitment.
See the deliverable
A report a board can act on
This is the actual output of an iDharma audit: a clear read your board, a partner bank, or a regulator can rely on - where every finding carries a decision, not just a description.
Inside every report
- Findings ranked by severity, worst first
- Each one mapped to the standards that apply to you
- A plain-English executive summary
- A prioritised fix list, signed by a named auditor
Sample Illustrative report built for a fictional company - every tier below produces a report in this format.
What we audit against
Measured against 31 named frameworks
Not a house scorecard. Every finding is written against a clause of a published framework, in the order those frameworks bind you.
What you have to protect
Data-protection law travels with the training data and the inference logs. These are the regimes that decide what your AI may hold, on what basis, and for how long.
- Privacy & data protection 9
What you choose to meet
The standards nobody forces on you and every procurement team asks about. Alignment here is what turns an audit into an answer for a buyer.
- AI standards & frameworks 4
- Security & resilience 6
The AI Audit
Choose the depth your AI needs
Quick scan
A fast, focused review of your core AI system, with the top risks surfaced.
$5,000
Request an audit
Compliance
Recommended
A deeper audit mapped to the standards and regulations you answer to.
$15,000
Request an audit
Risk
Our most thorough review, including threat scenarios and a full roadmap.
$25,000
Request an audit
Transparent, fixed-scope pricing - scoped with you before anything is charged. No payment until you approve the scope.
We show our work
Evidence over promises
An audit is only worth the record it leaves behind. Every engagement ends in a document you can hand to a regulator, a board or a buyer - findings, the standard each one touches, and the methodology we used, published and openly dated.
Every gap mapped to- NIST AI RMF
- ISO/IEC 42001
- EU AI Act
- SOC 2 & more
-
01
Standards
Mapped to which rule? NIST AI RMF, ISO/IEC 42001, EU AI Act, SOC 2 and more. -
02
Evidence
What was actually checked? Findings cited to what we examined, not to a self-graded checklist. -
03
Methodology
How was it decided? Versioned and openly dated, so you can see it stays current. -
04
Remediation
What gets fixed first? A prioritised remediation roadmap, not a generic score. -
05
Reporting
Can a board read it? Written to be read by the people who have to sign it off.
How we work
Four principles behind every audit
Independence
- We audit AI we didn’t build
- No vendor relationship, no conflict of interest
- An honest third-party read every time
Standards-based
- Every finding maps to NIST AI RMF, ISO/IEC 42001, or the EU AI Act
- Cited, not proprietary
- So results stand up to scrutiny
Named auditor
- A named reviewer signs every report
- You know who is accountable
- Not an anonymous model or faceless team
Speed without shortcuts
- Methodology updated in days not quarters
- Scope approved before billing starts
- Fast delivery with no compromise on rigour
Our mission
Independent proof your AI holds up
AI is being deployed faster than anyone can verify it. iDharma gives you the evidence that yours is accurate, fair, secure and compliant.
Request an AI audit
Evidence you can trust.
Proof you can defend.
Every audit maps to
- EU AI Act
- ISO/IEC 42001
- NIST AI RMF
- SOC 2
- HIPAA
- India DPDP
AI audit report
Illustrative sample
Findings by severity
- Fairness
- Security
- Privacy
- Reliability
- Transparency
Compliance mapped
- EUAI Act
- ISO42001
- NISTAI RMF
- SOC 2Type II
-
Sleep at night on AI risk
Documented evidence that your AI was independently checked.
-
Prove compliance, fix gaps
Every gap mapped to a regulation, with a prioritised fix roadmap.
-
Unblock the deal
An attestation that clears procurement gates and board scrutiny.
-
From risk to proof.
From audit to advantage.
Our vision
Checked once, then watched continuously - before the regulator or your board asks.
Every report is signed by a named auditor, and the methodology behind it is published and openly dated.
Brijesh Patel
Founder & Lead Auditor
Connect on LinkedIn
-
Named reviewer
on every audit -
No vendor
relationship
6
Frameworks mapped
NIST AI RMF, ISO/IEC 42001, EU AI Act, HIPAA, SOC 2, India DPDP
1–4wks
Typical engagement
Scope to report
From $5,000
Fixed fee
Scoped before you are charged
An aim, not a claim about today.
Continuous watching is the standard this firm is built toward.
-
Independent Assessment We audit AI we did not build.
-
Regulation Mapped Findings trace to a named framework.
-
Enterprise Ready Nothing charged until you approve the scope.
-
Evidence Based Conclusions tie to evidence you can inspect.
Ready to know where you stand?
Where your AI holds up on accuracy, fairness, security and compliance - and what to fix first.
Not sure where to start? Email us and we’ll recommend the right audit - or review a sample report.
NDA as standard
1–4 week delivery
Fixed-fee pricing
Expert-led, not automated
From Insights
Start with the writing
What Is an AI Audit? Scope, Standards, and What You Get
An independent review of what your AI actually does, measured against a named standard — not a certificate, and not a review of what the documentation says it does.
Read the notes
STANDARDS EXPLAINED
What an AI Governance Framework Actually Contains
Five working parts, not a policy document. What each one has to do, how to tell whether yours is real, and why a framework is not the same thing as compliance.
METHODOLOGY NOTES
How Much Does an AI Audit Cost? What Actually Drives the Number
Fixed prices, published turnarounds, and an honest account of the four things that decide which tier a system belongs in — including when the answer is "none of them yet".
INSIGHTS
Startups, Meet Your AI Stack: Budget‑Friendly Tools That Scale
For early-stage founders, building an AI-powered toolkit doesn’t have to break the bank. From ideation to growth mode, here’s how startups can tap into affordable, effective AI tools to autom
