The agentic shift: AI’s next phase

AI continues to attract sustained investment in Canada despite economic uncertainty. A significant majority of Canadian organizations (74 per cent) say AI will remain a top investment priority even if a recession occurs, aligning directly with global findings. At the same time, 65 per cent say they will continue investing in AI regardless of whether they can currently measure tangible return on investment.

Measurable outcomes do continue to lag intention. While 70 per cent of Canadian organizations say AI is already delivering meaningful business value, only three per cent report having achieved established, measurable returns on their AI investments. Globally, eight per cent report established AI ROI, underscoring a persistent gap between perceived value and provable outcomes.

Taken together, the findings suggest that while Canadian organizations are investing with conviction, most are still working to move from widespread experimentation and early use cases to repeatable results across multiple workflows.

Meanwhile, agentic AI—technologies capable of planning and taking independent, data-driven action—is no longer theoretical in Canada. Organizations are actively embedding AI agents into core business workflows, marking a clear shift in how work is organized and delivered.

Survey results show:

• 77 per cent of Canadian organizations surveyed are already using AI agents to assist with tasks such as knowledge sharing across teams and functions
• 66 per cent say they are moving toward a fully integrated human and AI workforce, where people and agents routinely work together.

Canadian organizations are also engaging with AI agents at different stages of maturity:

• 37 per cent are exploring or piloting AI agents
• 43 per cent are deploying, scaling or orchestrating agents across multiple workflows
• 20 per cent are developing or implementing multi‑agent systems.

Agent deployment is concentrated in foundational enterprise functions, including technology and IT, operations, risk and compliance, finance, and marketing and sales—closely mirroring global adoption patterns.

Canadian organizations are adopting AI agents at rates similar to global peers, but adoption alone is not determining outcomes. Where agents are embedded consistently into workflows and operating routines, organizations are beginning to see repeatable value. Where use remains fragmented or siloed, governance challenges, integration gaps and workforce resistance continue to limit impact.

When it comes to demonstrating return on investment, respondents cite several barriers that make it difficult to clearly measure or attribute value from AI initiatives. The most frequently cited include:

• Workforce skills gaps (43 per cent)
• Data privacy and cybersecurity risks (42 per cent)
• Difficulty quantifying indirect or long‑term benefits (40 per cent)

These factors point to challenges in governance, data access and accountability that complicate efforts to prove impact—even where AI is already in use.

Risk also shows up earlier in the lifecycle, shaping how quickly organizations move forward. Sixty‑six per cent of respondents say data security, privacy or risk concerns are actively slowing or pausing AI implementation, signaling caution around deployment, scale and use‑case expansion.

Viewed together, the findings suggest that risk‑related concerns are constraining AI in two distinct ways: they are making returns harder to measure and, separately, slowing the pace at which AI initiatives move from pilots to broader execution.

This pattern mirrors global findings, which show that while AI adoption continues to accelerate, only a small cohort of organizations is consistently translating AI—particularly agentic AI—into repeatable, organization‑wide performance gains. Access to technology is no longer the primary constraint; discipline in execution is.

Employee response to agentic AI varies widely across Canadian organizations. While 40 per cent report some level of adoption, a notable 31 per cent report employee resistance to AI agents, compared with 16 per cent globally.

The reasons employees are hesitant to use AI agents are consistently human‑centred:

• Trust, ethical or transparency concerns (51 per cent)
• Concerns about job security (39 per cent)
• Lack of confidence in AI skills and capabilities (38 per cent)

At the same time, agentic AI is already reshaping how organizations think about talent: 59 per cent say AI agents have changed how they hire entry‑level employees while 63 per cent say the same for experienced hires.

Looking ahead, many organizations expect AI agents to influence how performance is assessed and how roles evolve:

• 39 per cent expect performance metrics to place greater emphasis on AI‑collaboration skills
• 39 per cent expect more focus on human capabilities over tasks now handled by AI
• 36 per cent expect promotion criteria to prioritize AI literacy and effective agent delegation.

Compared with global peers, Canadian organizations are encountering greater workforce friction as agentic AI scales, making trust, enablement and leadership capability central to adoption.

AI is now firmly on the board agenda in Canada. Most organizations (80 per cent) report that boards regularly cover AI‑related topics, including cybersecurity, workforce impact and risk management.

However, only 63 per cent of leaders say they currently have the capabilities and governance in place to manage AI risks as the technology scales.

To manage risk, organizations are increasingly:

• Using human‑in‑the‑loop models to validate AI outputs (41 per cent)
• Restricting autonomous decision‑making in high‑risk use cases such as hiring (38 per cent)
• Embedding controls, monitoring and evaluation processes into agent workflows (36 per cent)

Global findings reinforce this pattern: organizations beginning to realize AI value at scale treat governance not as a compliance overlay, but as a foundational enabler of trust, coordination and accountability.