The decision ledger
Records every decision, instruction, context and output in real time in an immutable ledger, with millisecond precision.
Illustrative interface
AI governance platform
The governance and verification engine that turns every AI-driven decision into traceable, auditable, compliant evidence. Built for environments where mistakes are costly.
Decision record
DEC-2041 · Customer credit line
Illustrative example
AI models create value, but without an immutable record of how and why they reached a decision, their use in critical domains stays blocked by legal, regulatory and operational risk.
Decision accountability
When a model declines a credit line, flags a medical test or executes a transaction, your legal team must be able to show how and why. Without a verifiable trail, the decision cannot be defended.
Regulatory obligations and penalties
For high-risk systems, the EU AI Act requires continuous record-keeping, monitoring and human oversight (Articles 12 and 14). The most severe penalties reach €35 million or 7% of global annual turnover.
Unchecked drift
Models change with every update and parameter tweak. Without deterministic guardrails, small variations can cause material operational errors that go unnoticed.
IRIS Core doesn’t modify your models. It wraps any AI model in a verification layer, making every decision it makes auditable and provable.
Every decision leaves proof.
The input, context, instructions and output of every decision are recorded in real time with a cryptographic fingerprint, in an append-only ledger that cannot be altered.
Regulations become controls.
Native mapping to the leading AI frameworks: EU AI Act (Articles 9, 12, 14, 15), ISO/IEC 42001, NIST AI RMF and industry standards.
People decide when it matters.
Programmable checkpoints let domain experts review, correct or approve critical outputs before they take effect.
Independent yet integrated modules: enable them as you need them, and they all work on the same decision ledger.
Records every decision, instruction, context and output in real time in an immutable ledger, with millisecond precision.
Illustrative interface
Programmable rules for compliance, ethical constraints and safety, enforced before and after every model call.
rule: human_oversight_art14
if risk > 0.72 or confidence < 0.85
then route to review · risk analysts
else approve and sign
Illustrative interface
Compiles technical, contextual and operational evidence into dossiers readable by people and machines, for internal audit and regulators.
Illustrative interface
Visual exploration of past decisions: anomalies, alternative scenarios and reconstruction of reasoning paths.
Illustrative interface
IRIS Core sits along the path of every decision: it checks before, verifies after and records every step. Your systems and models stay exactly as they are.
Request
The request originates in your systems (API, ERP, service).
Rule check
Company policies and regulations are checked before the model is called; context is recorded.
Processing
Any model: proprietary, on-premises or cloud. IRIS never modifies it.
Output verification
Consistency, sources and thresholds are checked; signed evidence goes to the ledger.
Verified response
The decision arrives with its proof, ready for audit and review.
Any model
Proprietary, on-premises or cloud models.
Legacy systems too
Runs alongside ERP and AS/400 with no automated writes.
Immutable ledger
Every step produces verifiable evidence.
Industries where compliance is non-negotiable, errors carry legal consequences and black-box automation is simply not acceptable.
Financial services
Credit, fraud prevention, automated operations
Credit assessments, anti-money-laundering checks and operational decisions with explanations that hold up under regulatory inspection.
Reference use case
Explaining a declined credit application with documented alternatives.
SR 11-7 · Basel III
Healthcare and life sciences
Clinical protocols and diagnostic support
Diagnostic support and triage, with a clear, always-visible link between clinical guidelines and each recommendation.
Reference use case
Linking every recommendation to its underlying scientific sources.
EU MDR · FDA AI/ML guidance
Public sector and legal
Administrative decisions and regulatory summaries
Administrative determinations and regulatory summaries with precise citations and clear explanations for citizens.
Reference use case
Legal summaries with citations verified down to article and paragraph.
AI Act · high-risk categories
Enterprise governance
Visibility across every model in use
A single view for the board, general counsel and risk committees: models, vendors, drift and policy violations.
Reference use case
A compliance dashboard covering drift, costs and violations.
ISO/IEC 42001 · NIST AI RMF
IRIS at work · vertical applications
Technical and compliance documentation for IT, legal and risk management teams.
IRIS Core architecture
The decision ledger, boundary-verification algorithms and performance guarantees.
Request the whitepaperFor developersAPIs and SDKs
Documentation for the IRIS gateway, Python and TypeScript SDKs, and the rule syntax.
Go to the documentationComplianceEU AI Act matrix
An article-by-article mapping between IRIS Core capabilities and AI Act obligations (Articles 12, 14, 15).
Request the matrix