Introduction
Hundreds of enterprise leaders and technical experts gathered at VB Transform 2026 to discuss the use of generative AI agents in driving business outcomes.
Vice President Rachad Alao
Rachad Alao, vice president of product engineering at Cohere, a Canadian enterprise AI startup, shared insights on building agentic systems while maintaining data security and infrastructure control. Alao highlighted the importance of tight control over where sensitive data resides.
Defining AI Sovereignty
Alao argued that AI sovereignty involves more than simply downloading an open model or running applications behind a corporate firewall. It requires meticulous control over where data is stored and operated within jurisdictions an organization understands or directly controls.
Key Takeaways:
- Data residency: Ensuring data remains in controlled environments
- Jurisdictional control: Operating AI systems within familiar or direct control areas
- Infrastructure management: Managing GPUs and private cloud infrastructure effectively
- Governance systems: Efficiently routing requests among models
The Agent Stack
The discussion also covered the need to manage the full agent stack, from GPUs through to governance systems and connectors. This ensures that enterprises can leverage AI without compromising their sovereignty.
Steps for Achieving AI Sovereignty:
- Identify critical data and its location
- Implement tight control measures on where AI operates
- Ensure infrastructure is managed within controlled jurisdictions
- Develop governance systems for effective model routing
Conclusion
The conference highlighted the importance of enterprise leaders in maintaining sovereignty over their AI initiatives. By controlling the full agent stack, enterprises can ensure data security and maintain control over their infrastructure.