Every question
before the audit call.

When a denied loan, flagged transaction, or contested recommendation gets disputed — by a customer, regulator, or in court — your legal team needs to reconstruct exactly what the AI decided and why. Without an audit trail, that's weeks of forensic work, engineering time, and exposure. With Rubric, it's a single API call. Every decision is timestamped, signed, and retrievable in seconds.

You send them a link. Rubric generates court-admissible compliance reports on demand — SHA3-256 hashed, cryptographically signed, covering every AI decision in scope. What used to take weeks of preparation takes thirty seconds. You can see a sample compliance report here.

Boards are asking this question more often, and expecting more than a policy document in response. Rubric gives your leadership team a concrete, auditable answer — every AI decision your systems make is cryptographically recorded, independently verifiable, and reportable on demand. That's not a promise. It's a proof.

Several deadlines are converging at once. The EU AI Act becomes fully applicable in August 2026 — covering 14 articles across risk management, technical documentation, record-keeping, human oversight, and post-market monitoring for high-risk AI systems. In the US, Texas TRAIGA took effect January 2026, Colorado's AI Act takes effect June 2026, California has multiple AI transparency laws active now. Industry regulators — SR 11-7, FINRA 3110, FFIEC, ECOA, and EEOC — are already asking for AI decision documentation in examinations. Rubric covers all of it from a single SDK integration. The window to get ahead of this is closing.

The Trump administration is pushing for a unified federal standard and has challenged some state laws — but those state laws remain fully enforceable today. Only Congress or the courts can overturn them, and neither has. Colorado, California, and Texas laws are active and being enforced right now.

Industry-specific federal requirements — SR 11-7 for banks, FINRA 3110 for broker-dealers, FFIEC for financial institutions — have never been deregulated and aren't changing. Companies that wait for federal clarity are taking a real compliance risk.

The EU AI Act applies to any company whose AI affects EU citizens — regardless of where the company is based. If any of your customers, users, or counterparties are in Europe, you're in scope.

US state and industry-specific requirements apply independently of the EU entirely.

Rubric automatically records every decision your AI systems make — what the model decided, when, and why — and makes that record cryptographically signed, tamper-evident, and permanently verifiable. One line of code at startup. No new infrastructure required.

Every Rubric record includes a cryptographic signature, a precise timestamp, and an immutable anchor on Hedera's public ledger. Together these establish an unbroken chain of custody — proving the record existed at a specific moment and has not been altered since. The compliance report includes a SHA3-256 document hash as a verifiable sidecar.

This is the same structure used in digital forensics and e-discovery. It is designed to withstand legal scrutiny, not just pass an internal audit.

Rubric supports 13 frameworks out of the box: LangChain, LangGraph, AutoGen, LlamaIndex, CrewAI, Haystack, Semantic Kernel, Pydantic AI, OpenAI Agents SDK, Google ADK, Strands, DSPy, and the standard OpenAI client. One line instruments at the class level — no manual wrapping required. If you're building with AI today, you're almost certainly covered.

Yes. Rubric is built for exactly that. We've stress tested at over 3,200 requests per second with zero errors, across a five-node global federation. Our Merkle compression runs at 1,000,000:1 — meaning millions of decisions create minimal ledger footprint with sub-50μs signing latency.

Database logs are mutable. Anyone with access can alter or delete them. Rubric creates cryptographically signed records that can be independently verified by any third party — auditor, regulator, or court — without asking Rubric to confirm anything. That's a fundamentally different standard of proof.

You could. It typically takes 6–12 months, requires post-quantum cryptography expertise, distributed ledger infrastructure, and ongoing maintenance. Rubric is one line of code and operational in 30 minutes. Your engineers have higher-value problems to solve.

Your data never touches a public chain. Rubric stores only a cryptographic hash on the ledger — a mathematical fingerprint that proves the record existed and hasn't changed. The actual decision data is encrypted with AES-256-GCM, and your organization holds the encryption key. Rubric never has access to your plaintext data.

Every record is anchored to Hedera's public ledger — a decentralized network not controlled by Rubric. Even if Rubric ceased to exist tomorrow, every attestation remains independently verifiable forever. You own your records.

A formal security audit with Halborn is currently underway. Our pre-audit readiness assessment scored 97/100, with zero critical or high findings expected. We will publish the full report upon completion.

Thirty minutes. One line of code. No integration project, no professional services engagement, no new infrastructure to provision.

Both. Engineering installs it in thirty minutes and never thinks about it again. Compliance gets real-time dashboards, on-demand audit reports, and documentation that satisfies EU AI Act (all applicable articles), SR 11-7, NIST AI RMF, TX TRAIGA, CO AI Act, FINRA 3110, and FFIEC examiners. One integration. Two teams covered.

We're currently offering a 60-day free pilot to a small number of design partners. After that, pricing starts at $999/month for standard access, scaling to enterprise and dedicated tiers for high-volume or regulated-industry deployments. Less than one hour of outside counsel when the auditor calls.