Aegis is an enterprise governance layer for agentic AI systems, built in partnership with Kinaxis. It sits above existing AI tools such as Copilot, Claude, ChatGPT and enforces organizational policies at every agent action in real time. The system is grounded in several frameworks: OWASP Top 10 for LLM Applications, OWASP Agentic-AI threats, MITRE ATLAS and the NIST AI Risk Management Framework.
Every model call, tool call, memory read, and write-back passes through a single Policy Decision Point that enforces four levels of governance simultaneously: org-level invariants nobody can override, team-level policies, role-based permissions, and individual preferences. This is the L4 Hierarchical Values Cascade. Aegis's novel research contribution resolves at request time in under 100ms, with every decision recorded in a tamper-evident, hash-chained audit ledger.
The platform includes a live React admin console showing real-time PDP decisions, audit chain integrity, tenant cascade visualization, FinOps cost tracking, and policy exploration. Built with Python, FastAPI, PostgreSQL with row-level security, OPA policy-as-code, Keycloak OIDC, and pgvector for tenant-isolated RAG memory.
Five design principles enforced in code:
1. Authorization is data, not code
2. Identity never trusted from the token
3. Defence in depth (two independent isolation layers)
4. Fail-closed (default decision is DENY)
5. Everything audited immutably with hash chaining.
Ali Khreis Rahid Rishat Rafel
Software/AI Engineer Business Analyst
Ayegbe Jean-Noel Djahoua Toluwase Akinyemi
Software Engineer Project Coordinator
May H Elbanna Xujia Fan
UX Designer Software Engineer
Client: Emma Franks
Director, Enterprise Architecture - Kinaxis
Role in project: Emma Franks provided strategic guidance, oversight and expertise in governance architecture of enterprise agentic systems. She defined project goals, deliverables and provided key feedback throughout the project.
Technical Advisor: Professor Colin Bellinger
Associate Professor | Professeur Agrégé
School of Electrical Engineering and Computer Science | École de Science Informatique et de Génie Électrique
University of Ottawa | Université d'Ottawa
Role in project: Prof. Colin offered academic guidance in governance architecture design and testing of enterprise agentic systems.
This course offered a valuable combination of conceptual grounding and applied experience for every member of our team. The lectures clarified how engineering design, requirements gathering, and validation connect into a single process, and they gave us a shared vocabulary for reasoning about problem statements, user needs, and iterative prototyping instead of moving straight into coding or implementation details. That framing mattered a great deal on a project like ours, where the value of the system is not a single visible feature but a set of guarantees that had to be argued for.
Working on the Aegis AI Governance Platform gave us hands-on practice with topics that are usually only discussed in theory, such as scoping a minimum viable product, designing for specific user roles, and balancing new capabilities against the need for robustness, security, and testing. Building a governed platform for enterprise AI agents pushed us to translate abstract governance needs into concrete, testable behaviour: per-action policy decisions, tenant isolation, a hierarchical values cascade, PII redaction, cost controls, and a tamper-evident audit trail. We experienced directly how client feedback can reshape both the interface and the underlying design: guidance from our sponsor, Kinaxis, led us to seed the assistant with their public corporate values, to make every governance claim reducible to an exact, machine-checkable test, and to harden the system for a stable, externally testable deployment. The project also forced us to coordinate frontend, backend, policy, and data work across several people, which strengthened our teamwork, communication, and version-control habits.
As a result, the course has advanced our academic and professional development in several ways. It reinforced the importance of clear documentation and explicit design reasoning which will serve us well in future research and capstone work. It also gave us a concrete, real system we can describe in portfolios and interviews, demonstrating experience with identity and access control, multi-tenant isolation, policy-as-data, privacy and data-residency controls, auditability, and cloud deployment, all delivered in collaboration with an external industry partner. Overall, the combination of lectures and project work has made us more confident about taking on complex, multi-stakeholder software projects in our future studies and careers.