Institutional AI
Sovereign AI for Ontario institutions that can't compromise on data
For Ontario public-sector bodies, hospitals, universities and Toronto enterprises, we design AI platforms that keep data in Canada, enforce tenant isolation at the database level, and pass security review, delivered through a structured four-stage program.
- Data residency
- Canada
- Database-level isolation
- RLS
- Stage engagement model
- 4
- Full action logging
- Audit
Capabilities
Enterprise-grade by architecture, not by promise
Postgres row-level security
Multi-tenant isolation enforced inside the database, so one department can never query another's data, even through a bug.
Sovereign deployment
Private cloud in Canadian regions (Toronto, Montréal) or on-premise, with open-weight model options.
Role-based access
SSO integration and granular permissions mapped to your organisational structure.
Audit trails
Every prompt, retrieval and action logged for compliance and internal review.
Governance frameworks
AI use policies and risk assessments aligned to PIPEDA, PHIPA and FIPPA/MFIPPA where they apply.
Security review ready
Architecture documentation and threat models prepared for your IT security team.
Multi-tenant security
Isolation you can verify
Row-level security policies travel with the data. Each request carries a tenant context, and Postgres refuses to return any row outside it.
ALTER TABLE documents ENABLE ROW LEVEL SECURITY;
CREATE POLICY tenant_isolation ON documents
USING (tenant_id = current_setting('app.tenant_id')::uuid);
-- Every query from the AI layer is scoped automatically:
SET app.tenant_id = '…'; -- set per request from the verified sessionEngagement framework
Four stages from strategy to scale
Each stage ends with a decision gate, so you commit budget only as value is proven.
- 01
Assess
Readiness review across data, infrastructure, policy and people. Use cases ranked by value and risk.
- 02
Architect
Sovereign reference architecture, security model, governance framework and business case.
- 03
Pilot
One high-value use case built on production-grade infrastructure and measured against agreed KPIs.
- 04
Scale
Roll out across departments with training, change management and an internal centre of excellence.
Reference stack
- PostgreSQL + RLS
- pgvector
- Self-hosted LLMs
- n8n
- Keycloak / Entra ID SSO
- Kubernetes / Docker
FAQ
Questions we hear most
What does sovereign AI mean?
Sovereign AI means your data, models and infrastructure stay under your jurisdiction and control, hosted in Canada, with no dependency on sending sensitive data to third-party services.
Why row-level security instead of separate databases?
Postgres row-level security enforces tenant isolation inside the database engine itself, giving strong guarantees while keeping one maintainable platform. Separate databases remain an option for the most sensitive tenants.
Can we start small?
Yes. The Assess stage is a fixed-scope engagement, and nothing beyond it is committed until you've seen the findings and business case.
Start with a confidential readiness assessment
A fixed-scope review of where AI can create value in your institution, and what it takes to do safely.