Cryptographic runtime attestation for production AI.
Tamper-evident audit trails. Per-decision cryptographic evidence. Built for organizations deploying AI in regulated environments and high-stakes decision contexts.
Deployment
Cloud-native AWS
Audit Trail
Tamper-evident chain
The evidence gap
Your AI made ten thousand decisions today. Prove what any one of them was.
Logs can be edited, deleted, or simply never written. When a regulator, a customer, or a court asks what your AI actually did, a log line is an assertion — not evidence. LogiRoot issues a signed, tamper-evident decision receipt for every governed action, at the moment it happens.
Audit without archaeology
Audit prep shouldn't be a dig through last quarter's logs.
With LogiRoot, evidence assembles itself as a side effect of running. Every decision joins a hash-linked chain that can be handed to an auditor.
Enforcement, not observation
Policies that only watch are not policies.
Dashboards tell you what went wrong after it shipped. LogiRoot sits in the request path and enforces policy before the action executes — allow or block — and every outcome carries its own decision receipt.
How it works
The flight recorder for production AI.
01
Request
Your application asks before the AI acts — one REST call into the governance layer, in the request path.
02
Decision
Policy evaluation answers in the request path: allow or block. Enforcement happens before the action executes.
03
Decision receipt
A cryptographically signed, hash-linked receipt records the decision — tamper-evident.
The Platform
Decision-level evidence, verifiable at runtime.
Drop-in middleware between your AI provider and your application. Every decision receipted. Every output governed.
Cryptographic attestation
Every model output carries a cryptographic receipt — who requested it, what model responded, what governance applied.
RUNTIME
Tamper-evident audit trail
Append-only chain with sequentially linked entries. Exportable to support record-keeping and audit-retention requirements.
PROVABLE
Runtime policy enforcement
Pre-execution policy checks run before any output ships. Define governance boundaries in code; the sidecar enforces them on every decision.
DECISION-TIME
Multi-model governance
Drop-in middleware. Works with OpenAI, Anthropic, Mistral, Llama, or any OpenAI-compatible endpoint. One governance surface across providers.
UNIVERSAL
Cloud-native deployment
Runs on AWS container compute with KMS-backed secrets via AWS Secrets Manager. Integrates through a single REST API — no application rewrites.
CLOUD-NATIVE
Pricing
Subscribe through AWS Marketplace.
Early-access pricing for founding customers — monthly and annual contracts. Custom and multi-tenant configurations via AWS Marketplace Private Offer.
Available where you procure.
Cloud-Native
Deployed where your AI already lives.
Native integrations with AWS services. Sidecar deployment, no application rewrites.
AWS Bedrock
Governs Bedrock-hosted model calls through the API — a signed receipt on every governed call, no model changes required.
Amazon SageMaker
Model-agnostic governance covers SageMaker-hosted endpoints alongside any other provider, through one API. No retraining or redeployment.
AWS CloudTrail
Every governance decision is recorded in a tamper-evident, exportable evidence trail for your audit and SIEM pipelines.
AWS IAM
Runs on AWS IAM-scoped infrastructure; every signed receipt carries per-tenant attribution. No parallel identity store to maintain.
Secrets Manager
Signing material custodied via Secrets Manager with KMS-backed encryption. Customer-managed keys optional.
ECS / EKS / EC2
Runs on AWS container compute and integrates through a single REST API. No application rewrites.
Standards & Regulatory Alignment
What NIST published in June 2026.
NIST published a mathematical proof — an extension of Gödel's incompleteness theorems to AI security — showing that no finite set of guardrails is universally robust against adversarial prompts. Point-in-time defenses have inherent limits. NIST's prescription is a continuous one: ongoing red-team discovery, continuous updates, and operational resilience. LogiRoot is built for the evidence half of that model. It does not claim completeness — it produces a durable record of every governance decision, so you can show what was enforced, when, and under which policy.
NIST, June 2026 — “Mathematical Proof Supports Transition to Continuous Monitor-and-Update.”
Evidence aligned to
EU AI Act
Articles 12 & 26 — record-keeping & deployer logging
Designed to help deployers meet the record-keeping and log-retention duties of Articles 12 and 26 — tamper-evident records you can retain and produce on demand.
NIST AI RMF
Govern · Map · Measure — continuous-update model
Maps to the Govern, Map, and Measure functions with signed, durable records of each decision — the continuous-monitor-and-update model NIST now prescribes.
ISO/IEC 42001
AI management system — operational evidence
Supports an AI management system by enforcing governance policy in-path and producing durable, cryptographically signed records of each decision.
SOC 2
Customer-side audit evidence
Evidence for your own SOC 2 audit — signed records of every APPROVE and REJECT.
LogiRoot produces runtime evidence aligned to the approach these frameworks prescribe. It holds no certification from, and no endorsement by, NIST, the EU, ISO, or the AICPA — conformity remains your organization's process. LogiRoot makes the evidence part automatic.
Available on AWS Marketplace.
LogiRoot is publicly available on AWS Marketplace today — subscribe directly and procure under your existing AWS spend. Custom and multi-tenant configurations are available via AWS Marketplace Private Offer.
Founding customers & private offers
LogiRoot is publicly available on AWS Marketplace — subscribe directly to get started. We also offer custom or multi-tenant configurations via AWS Marketplace Private Offer. Tell us what you're deploying — every request is reviewed by a human — or reach us at support@logirootai.com.