Canadian Government Leaders Confront Reality of AI Deployment
Government agencies across Canada are moving past AI experiments and into operational reality. The shift came into sharp focus at the OpenText Government Summit in Ottawa, where public sector leaders, technology executives, and cybersecurity experts gathered to discuss how to operationalize AI responsibly while protecting sensitive information and maintaining public trust.
The conversation has fundamentally changed. Governments no longer debate whether AI will transform public sector operations. They now focus on how to deploy it securely and at scale.
Data Quality Determines AI Success or Failure
The summit's central insight was direct: AI systems fail when they operate on poor data.
OpenText CEO Ayman Antoun stated that 85% of AI projects have failed, mostly because of underlying data quality. "Chief data officers don't trust their own data," Antoun said. "If you don't have a good view of your data management framework, don't invest in AI. You won't like the outcome."
Many agencies have invested in data platforms, analytics tools, and AI copilots. But critical information remains scattered across siloed systems, unstructured documents, and disconnected workflows. AI cannot deliver trustworthy results from incomplete or poorly governed information.
Antoun offered three questions agencies should answer before deploying AI:
- Does it serve citizens better?
- Does it make government employees work more effectively?
- Does it improve the products we use?
"Fight the urge to invest in AI for the sake of AI," Antoun said.
One concrete example: OpenText's National Operations Center stores information for 300,000 clients. AI agents there work alongside cybersecurity experts to detect threats more efficiently, filtering out false alarms from genuine risks.
Sovereignty Becomes Operational Priority
Governments are increasingly prioritizing data sovereignty, secure cloud architectures, and vendor flexibility as AI deployment accelerates. Sovereign digital capabilities must support responsible AI adoption while protecting government information and strengthening operational resilience.
Mark Schaan, Canadian Associate Deputy Minister for Innovation, Science and Economic Development, emphasized collaboration between government and industry. "Sovereignty does not equal solitude," Schaan said.
He identified three areas where industry can partner with government: helping communicate positive technology stories to counter AI skepticism, identifying where AI should be deployed, and building relationships to shape the future together.
AWS and OpenText are collaborating on sovereign cloud architectures designed specifically for Government of Canada requirements. The partnership frames sovereign cloud not as infrastructure alone, but as foundational capability for what presenters called the "Sovereign Intelligence Era."
Privacy and Governance Cannot Be Added Later
As agencies move from pilots to production, privacy and compliance concerns have moved to the center of adoption strategies. Privacy and governance cannot be retrofitted after AI systems go live.
A session with leaders from the Office of the Privacy Commissioner of Canada emphasized that agencies need stronger visibility into how AI systems access and use information. Privacy and compliance must be embedded into workflows by design.
Identity governance presents a particular challenge in what officials called the "agentic era," where AI agents can provision access, execute actions, and interact across systems with limited human intervention. New questions emerge: Who owns AI agents? How are permissions managed? How are actions audited? What controls exist to shut systems down if needed?
Darcy Pierlot, Chief Technology Officer for Shared Services Canada, stressed the stakes. "I deliver solutions and services across the Government of Canada," Pierlot said. "Information around passports, immigration, retirement, social services - we house all of that data. We need to deploy systems that are still relevant and useful but also secure and responsible."
The summit included a demonstration of OpenText Aviator Studio, where government agencies can create AI agents to automate tasks specific to public sector use cases. With proper permissions, these agents can cross boundaries between SAP, Microsoft, and OpenText systems to accomplish complex processes.
Unstructured Content Is Critical Infrastructure
Many agencies underestimate the value of unstructured information. Documents, emails, reports, multimedia files, and case records often contain the majority of institutional knowledge-but remain inaccessible to AI systems.
Unstructured content is no longer archived material. It is becoming primary input into AI-driven decision-making, operational workflows, and citizen services.
This shift has major implications for government modernization. AI readiness increasingly depends on content readiness. Governance and metadata become strategic capabilities. Enterprise search evolves into enterprise knowledge activation.
AI Must Integrate Into Operations
AI cannot succeed in isolation. AI initiatives must connect seamlessly into mission workflows, operational systems, service management environments, and security frameworks.
Sessions focused on AI orchestration demonstrated how AI agents can collaborate across systems to automate complex processes, adapt to changing conditions, and reduce manual effort. Without integrated operational frameworks, advanced AI capabilities risk becoming disconnected point solutions rather than enterprise-wide enablers.
The broader message was consistent: responsible AI is not just a technology challenge. It is an operational transformation challenge.
From Strategy to Execution
The strongest theme from the summit was the transition from principle to execution. Governments understand the importance of responsible AI, data governance, cybersecurity, and public trust. The next challenge is operationalizing those principles across complex, real-world environments.
That requires connecting data and content across systems, embedding governance into AI workflows, ensuring transparency and auditability, building secure sovereign digital foundations, and activating institutional knowledge at scale.
The path forward is not about replacing existing investments. It is about connecting them into a unified operational fabric capable of supporting trusted AI and modern digital government.
As AI adoption accelerates across the public sector, agencies that succeed will be those that combine innovation with governance, speed with trust, and automation with accountability. The future of digital government will be defined not by how much AI agencies deploy, but by how responsibly and effectively they operationalize it.
For government professionals exploring AI adoption, resources on AI for Government and Generative AI and LLM provide practical frameworks for responsible deployment.
Your membership also unlocks:
