OAIS — AI Sovereignty Infrastructure
AI Sovereignty Infrastructure

AI was never the problem.
Deploying it without
sovereignty was.

OAIS builds the infrastructure that makes AI yours — captured, blocked, governed, and optimised.

287 automated tests
97%+ branch coverage
April 2026 running since
0 chain gaps
OSFI E-23 Compliance Deadline
--- days
:
-- hours
:
-- min
May 1, 2027 — Mandatory for all federally regulated Canadian FIs

Two Products. One Chain.

Sentinel Core captures and anchors every AI interaction. Sentinel Guard intercepts them at the browser. Together, they form a cryptographic chain of custody for every AI interaction in your organisation.

Sentinel Core

Cryptographic AI Interaction Registry

One line of code. Every AI interaction is HMAC-hashed with a key only you hold, Ed25519-signed, Merkle-anchored, and timestamped by an independent RFC 3161 authority. OAIS sees the fingerprint. Never the content.

Layer 1

HMAC-SHA3-256 Hashing

Content hashed with a tenant key that OAIS never holds. Inputs and outputs become irreversible fingerprints.

Layer 2

Ed25519 Signing

Each record is signed by the client agent. Tamper-evident from the moment of capture.

Layer 3

RFC 6962 Merkle Anchoring

Records are batched into Merkle trees. Any single change invalidates the entire chain above it.

Layer 4

RFC 3161 Timestamping

An independent timestamp authority proves records existed at a specific time. No OAIS infrastructure access required to verify.

Independently Verifiable

Any third party can verify the entire chain with a standalone script. No OAIS account, no API access, no trust required.

Sentinel Guard

Browser Extension for Shadow AI Detection & DLP

Deployed via Chrome Enterprise Policy or Intune. Sentinel Guard detects AI tool usage across Chrome and Edge, enforces DLP rules before data leaves the browser, and feeds metadata records into the same Sentinel Core registry.

Shadow AI discovery across ChatGPT, Claude, Gemini, Perplexity
DLP enforcement at the browser layer — stop what you don't want leaving before it leaves
Metadata-only by default — content capture is opt-in
Same cryptographic chain as Sentinel Core — one registry, complete coverage

We are not here to make the
Large Liability Model more powerful.

We are here to make you sovereign over it.

01

Capture

Every AI interaction, in and out, hashed with a key only the client holds. OAIS sees the fingerprint, never the content.

02

Block

DLP enforcement at the browser and API layer. Stop what you don't want leaving before it leaves.

03

Govern

AI output that triggers autonomous action passes through a deterministic governance gate before it executes.

04

Optimise

The audit chain is data. Use it to optimise tokenomics, fine-tune models, build proprietary datasets, demonstrate ROI, streamline workflows.

OSFI E-23

Effective May 2027. Enterprise-wide AI model risk management for all federally regulated Canadian FIs.

EU AI Act

Mandatory compliance for AI systems in regulated contexts across European markets.

ISO 42001

International standard for AI management systems. Sentinel produces the evidence automatically.

The Audit Chain Is Data

Once every AI interaction is captured cryptographically by Sentinel Core, new capabilities emerge that were previously impossible.

Regulatory Evidence on Demand

OSFI E-23, EU AI Act, ISO 42001. Export a verifiable, timestamped chain for any audit window in seconds.

Shadow AI Discovery

Find every AI tool in use across your organisation. Know what's happening before a regulator asks.

Proprietary AI Datasets

Your interaction data, structured and labelled. Build proprietary datasets for fine-tuning and competitive advantage.

Token Optimisation

Cost per outcome, not cost per call. See which interactions produce value and which waste tokens.

Governed Autonomous Agents

AI that acts within your boundaries. Deterministic governance gates that are architecturally incapable of being bypassed.

Workflow Streamlining

Audit chain insights reveal bottlenecks, redundant interactions, and optimisation opportunities across your AI-augmented workflows.

Tailored to Your Environment

Every deployment is different. Pricing is scoped to your systems, data volume, integration requirements, and compliance objectives.

Pilot Program
Let’s Scope It

30 days. One AI system. Full Sentinel Core functionality with dedicated onboarding. We’ll define scope, integration approach, and success criteria together.

  • Full Sentinel Core functionality
  • Cryptographic chain of custody
  • RFC 3161 timestamping
  • Independent verification
  • Dedicated onboarding support
Start a Conversation
Sentinel Core
Custom

Priced by the number of AI systems monitored, interaction volume, and integration complexity. Includes platform licensing, implementation, and ongoing support. Consulting and advisory services available.

  • Full compliance-grade audit trail
  • Cryptographic chain of custody with RFC 3161
  • Real-time and batch integration options
  • OSFI-aligned governance reporting
Talk to Our Team
Sentinel Guard
Contact Us

Per-seat licensing for Chrome and Edge. Volume-based pricing scaled to your organization. Annual and multi-year options available.

  • Shadow AI discovery
  • DLP enforcement
  • Chrome Enterprise / Intune deployment
  • Records feed into Sentinel Core
Get a Quote

Every engagement starts with a scoping conversation. No commitment required.

Built by Practitioners,
Not Observers

Founder & Architect

Daniel Fruman

Designed and built the Sentinel architecture. Over a decade of hands-on experience in healthcare privacy compliance and data governance, building and enforcing governance systems in regulated Canadian environments.

Co-Founders

Founding Team

Three co-founders with senior experience spanning institutional investment management, insurance operations, and financial services. Combined expertise in enterprise governance, regulatory compliance, actuarial science, and risk modelling — drawn directly from the industries Sentinel is built to serve.

Built in Canada. Focused on OSFI E-23. Serving regulated enterprises globally.

Frequently Asked Questions

What is OAIS?

OAIS (Optimized Artificial Intelligence Systems Inc.) builds AI sovereignty infrastructure for regulated enterprises. Its two products — Sentinel Core and Sentinel Guard — capture, block, govern, and optimise every AI interaction in an organisation, creating a cryptographic chain of custody that is independently verifiable.

What is Sentinel Core?

Sentinel Core is a cryptographic AI interaction registry. With one line of code, every AI interaction is HMAC-hashed with a key only the client holds, Ed25519-signed, Merkle-anchored (RFC 6962), and timestamped by an independent RFC 3161 authority. OAIS sees the fingerprint, never the content. Any third party can verify the entire chain without an OAIS account.

What is Sentinel Guard?

Sentinel Guard is a browser extension for shadow AI detection and data loss prevention. Deployed via Chrome Enterprise Policy or Intune, it detects AI tool usage across Chrome and Edge, enforces DLP rules before data leaves the browser, and feeds metadata records into the Sentinel Core registry. It is metadata-only by default — content capture is opt-in.

How does OAIS help with OSFI E-23 compliance?

OSFI E-23, effective May 2027, requires enterprise-wide AI model risk management for all federally regulated Canadian financial institutions. Sentinel Core and Sentinel Guard automatically produce the cryptographic audit evidence needed to demonstrate control over AI interactions — including tamper-evident records, independent timestamps, and exportable compliance reports for any audit window.

Does OAIS see my data?

No. Sentinel Core hashes content with an HMAC key that only the client holds. OAIS sees the cryptographic fingerprint — never the content itself. Sentinel Guard is metadata-only by default; content capture is an opt-in configuration. The entire chain can be verified by any third party using a standalone script, without OAIS infrastructure access.

What compliance frameworks does OAIS support?

OAIS supports OSFI E-23 (mandatory May 2027 for Canadian federally regulated financial institutions), the EU AI Act (mandatory for AI systems in regulated European contexts), and ISO 42001 (the international standard for AI management systems). Sentinel produces the audit evidence automatically.

How is Sentinel Guard deployed?

Sentinel Guard is deployed via Chrome Enterprise Policy or Microsoft Intune. It runs as a browser extension on Chrome and Edge, requiring no changes to existing AI tools or workflows. It detects AI usage, enforces DLP rules at the browser layer, and feeds records into the same Sentinel Core cryptographic registry.

What does OAIS pricing look like?

Pricing is scoped to each deployment: number of AI systems monitored, interaction volume, integration complexity, and compliance objectives. Sentinel Core uses custom platform licensing. Sentinel Guard uses per-seat licensing with volume-based pricing. A 30-day pilot program is available. Every engagement starts with a scoping conversation.

Request a Demo

If your organisation is preparing for OSFI E-23 compliance, evaluating AI governance solutions, or needs to demonstrate control over AI interactions — we welcome a conversation.

Or reach us directly at info@oais.ai