At a glance
- AI report: Survey insights on AI readiness and adoption.
- The shift from AI assistants to agentic digital coworkers.
- The future of work, talent, and organizational design.
- Governance, trust, and responsible AI at scale.
- Preparing organizations for an AI-embedded future.
The era of the AI pilot is rapidly closing. What organizations once approached as isolated experiments has shifted into an operating reality, and as vendor capabilities converge, the advantage now lies less in model selection than in disciplined planning, execution, and adoption at scale.
The conversation in boardrooms across Canada is no longer about whether AI matters or how fast it is moving. We have now moved on to questioning whether organizations are building the operating models, governance structures, and human capabilities required to make AI actually matter at scale before the gap between what the technology enables and what their business can absorb becomes too wide to close.
We’ve seen a consistent pattern across hundreds of client engagements in industries including financial services, manufacturing, energy, real estate, and private equity. The organizations that go beyond just having the most sophisticated technology to focusing on connecting AI to measurable business outcomes, embedding governance, and investing in their people are the ones excelling in this era of tectonic change. By contrast, those that are stalling still treat AI as a technology project rather than an operating model shift.
An Angus Reid survey polling 520 Canadian business leaders makes the readiness gap more visible. Less than one in five leaders (18%) say they are actively embedding AI into workflows and operations, and when asked how AI will shape their organizations over the next four years, more than a quarter (27%) believe it will have minimal impact at all. The technology has arrived at a speed we have never seen before, but the operating discipline to harness it is trailing behind.
Our perspective, reflected in AI Vision 2030, is built on a clear conviction: the future belongs to organizations that harness AI to empower their people. When envisioning a human-led, AI-embedded world, success will come from putting people at the centre, grounding every initiative in outcomes, moving forward responsibly, and building capabilities designed to scale alongside the business and alongside the evolution of AI.
This report shares what we are seeing in the market, what we have learned from our own transformation, and what it takes to move from ambition to impact.
From assistant to operator: The agentic cowork shift
Strategic clarity starts with understanding where AI was, and where it is heading, not just where it is today. The distinction between generative AI and agentic AI is more than a technical nuance. It reshapes how work is structured, how teams are composed, and how value gets delivered.
Generative AI, which entered the mainstream in late 2022, operates as a capable assistant. It synthesizes information, drafts content, and answers questions based on the data it has been given. It is powerful, but fundamentally reactive, waiting for inputs and delivering outputs.
Agentic AI represents a different paradigm. These systems can reason, plan, and execute tasks across multiple platforms and workflows; and the contrast becomes clearer in practice. A generative tool writes a report when asked, where an agentic system monitors performance data, identifies an anomaly, drafts an analysis, routes it to the right stakeholder, and proposes a corrective action, all within established governance frameworks.
More recently, we are seeing the manifestation of agentic AI as digital coworkers, operating alongside professionals to handle workflows that were previously manual, time-intensive, or bottlenecked by talent availability.
Industry analysts project that by 2028, a third of enterprise software applications will include agentic capabilities, up from less than one percent in 2024. The agentic and cowork shifts are already underway, and they carry implications for every function: how finance teams close the books, how product and marketing teams tell the story, how engagement teams deliver advisory work, how organizations price and structure their services.
The shift is far from abstract at BDO. We believe that the best advisory comes from our own trial and error. BDO has moved past pilots and copilots and into deploying our own agentic orchestration layer across firm workflows, coordinating systems that scale insight and augment expertise rather than just assisting with isolated tasks. Our perspective on what works, what fails, and what scales at the agentic stage is grounded in that experience.
The immediate implication is significant. For functions that have spent the past decade getting better at reporting on what happened, agentic AI is now offering to do something about it. The bottleneck shifts from analysis to thought partnership to decision-making, which is exactly where executives have always wanted it to sit.
AI signals: What the market is telling us
Across our engagements, six interconnected forces are reshaping the operating environment for our clients. These are observable patterns rather than predictions, and each one carries a structural shift that business leaders need to be preparing for during this budget cycle, not the next one.
Teams have traditionally run the process and checked the outputs. That model is flipping as AI is beginning to run the process while people run the proof. Roles are evolving into hybrid models that shift human value toward judgment, trust, relationships, and strategic outcomes. For instance, to ensure effective “cowork outcomes”, business users are now expected to be the lead AI operators: defining the vision and work, supervising the workflow, and validating the output; this means a different way of thinking and of operating. Organizations that fail to redesign roles and workflows around this reality risk losing both talent and competitive positioning. AI may not reduce the overall need for professionals, but it will reduce the market value of certain task categories. As a result, organizations must redesign roles, leverage models, and rethink pricing structures before the economics are forced upon them.
AI capabilities are evolving faster than most organizations can absorb. Based on the new features released by major AI vendors in the past 4-5 months, we are in a world where there is a major AI release every month. Organizations who looked at AI as a technology solution and didn’t invest in effective planning and enablement are seeing lower ROI or limited adoption. The Angus Reid data confirms this directly: 46% of Canadian business leaders report experimenting with AI without yet achieving meaningful ROI. Returns on AI investment ultimately depends on the quality of a company's data foundation, core infrastructure, guardrails, training, internal communication, and cultural commitment behind the rollout.
No single AI platform will serve every need across the enterprise, and many organizations are already orchestrating multiple specialized systems. Vendor strategy, integration architecture, and interoperability have moved out of back-office IT and into core executive decision-making. As agents move from experimental tooling to productized capabilities embedded in core workflows, a new category of risk follows them.
We see it firsthand in client environments that have grown to apply their AI strategy across multiple aspects of their business, whether that be an intelligence layer on top of their Enterprise Resource Planning (ERP), an enterprise license for one foundation model, copilots bolted onto productivity tools, or a handful of departmental agents that nobody fully tracks. ‘AI sprawl,’ as it is sometimes called, multiplies governance overhead, integration cost, and reputational exposure faster than most organizations are noticing. Without a deliberate orchestration strategy, the sprawl itself becomes the bottleneck that the AI program was supposed to remove.
As AI handles more analytical and process-intensive work, traditional talent shortages may ease in some areas, but new gaps are emerging. The critical shortage is less in technology specialists, where the labour market is responsive, and more in AI-fluent professionals across finance, operations, HR, and client-facing roles who can design, supervise, validate, and scale intelligent systems within their own functions.
One useful way to think about this is an analogy to offshoring. Two decades ago, when professional services firms began routing work to teams in lower-cost markets, the conventional prediction was that onshore headcount would fall. It did not. Instead, the new capacity expanded the aperture of what firms could take on, and it made problem-solving more accessible for customers, and onshore teams grew alongside it. The same dynamic is playing out with AI capability now, and at a much faster pace. Organizations that prioritize AI adoption and build AI fluency deliberately throughout the workforce will hold a decisive advantage over those still waiting for the right hire to come along.
As AI takes on more consequential work, clients and regulators alike are choosing the partners and platforms that deliver accuracy with traceability, auditability, and human accountability built in.
The pattern is familiar from cybersecurity. SOC 2 and ISO 27001 went from differentiator to table stakes within a decade, and ISO 42001, the emerging international standard for AI management systems, appears to be following the same trajectory.
Mid-market organizations that don’t treat governance as an essential building block will find that the buyers and regulators they answer to have already moved on. Trust has shifted from a soft consideration into a buying criterion that determines which businesses make the shortlist.
Agentic AI is automating and commoditizing many of the workflows that the advisory industry has spent decades building careers around. The billable-hour model, which has been the foundation of how time, expertise, and value have been priced in many industries, does not survive that math intact.
Forward-looking firms, our own included, are already moving toward outcome-based pricing options that align fees to the value delivered for clients rather than the hours invested in delivering it.
From pilots to operating model transformations
What that shift looks like in practice is where most organizations stall. The ones moving from pilot to scaled deployment now share one trait, which is that they treat AI as a unified transformation across decisions, risk, talent, and operating structure rather than as a loose collection of separate initiatives looking for a tool to justify them. That means AI planning, managing use cases, data products, AI architecture, technology changes, risk controls, performance measurement, and adoption as one enterprise AI program.
The first shift is in how organizations manage their expanding AI capabilities. Agents that take on work previously performed by people need to be governed with the same rigour. This means building life cycle management practices, from design and deployment through monitoring, performance measurement, and retirement, with clear accountability at every stage. Ungoverned AI-enabled work introduces a category of risk that compounds faster than most legacy oversight structures were ever designed to catch.
Responsible AI practices belong at the outset and throughout the life cycle, not bolted on after adoption is underway. This is one of the most consistent misconceptions we encounter in client conversations, and one of the most expensive when it goes uncorrected. Many leaders treat AI and data governance as a brake on innovation when, in our experience, governance is actually what allows the accelerator to be pressed safely.
Organizations that invest in AI and data readiness, build clear operating models, policies around acceptable use, accountability structures, and alignment with emerging frameworks like ISO 42001 are not slowing themselves down; they are scaling with confidence, because every new use case clears a defined risk path rather than opening a new internal debate— low-risk productivity use cases should move quickly, while consequential, regulated, or customer-impacting decisions require deeper responsible-AI review. Governance, done well, is the backbone of sustainable AI-driven growth, and it earns the trust of clients, regulators, and the people inside the organization who are ultimately being asked to use technology in their day-to-day work.
The second shift needed to move from pilot to production is recognizing that AI fluency cannot remain concentrated within technology teams. When capability lives only inside IT or a digital innovation group, AI stays a departmental tool that scales only as far as that one department’s reach.
Many organizations spend a year or more building proofs of concept that scale exactly nowhere, because the people who need them most (in finance, operations, HR, and client-facing teams) were never part of designing them. When every professional has foundational fluency, the centre of gravity shifts, and solutions get co-created across functions. Emerging capabilities translate into real-world applications faster, and the organization gains the ability to respond to change as it happens rather than after the fact.
Most critically, an AI strategy must start with business problems and measurable outcomes, not with the technology itself. CFOs in particular have spent the past three years watching technology investment land on their balance sheets without commensurate ROI, in large part because too many AI initiatives have begun with a tool in search of a justification.
The organizations seeing real returns take the opposite approach: they define what success looks like in business terms, identify where value can be created, and then apply AI alongside data, process change, and people development. This outcome-first approach separates lasting transformation from expensive experimentation.
What we learned as client zero
The most credible guidance on AI transformation comes from lived experience rather than theoretical frameworks. Rather than advising clients from the sidelines, we chose to become our own first AI client, an experience we refer to as our client zero story. As a firm of more than 5,000 professionals serving organizations across every major industry in Canada, our transformation has been deliberate, and we have made many of the same mistakes other organizations are making now.
Each stage of our experience has reinforced the same four disciplines we now bring to every client engagement: start with the business problem, invest deliberately in people, govern responsibly from day one, and measure what actually matters to the business rather than vanity metrics about tool adoption.
Our journey is ongoing. We are now expanding our agentic orchestration layer more deeply across firm workflows, and as reflected in AI Vision 2030, the discipline of getting it right “at scale” matters more to us than the milestone of declaring it done.
Building the next era, together
The future belongs to organizations that approach AI with clarity and intent, deliberately aligning their operating model transformation, governance, and talent strategies with what the technology now makes possible. Leadership teams face a real choice: they can either approach AI adoption with the strategic and human-centred design required to make it last, or they can continue with fragmented pilots, unclear ownership, and tools that never connect to outcomes. The first path compounds advantage over time. The second compounds exposure to risk and competitive irrelevance, and the cost of changing course gets steeper the longer it’s postponed.
BDO’s AI Vision 2030 offers a different path, one that is human-led and AI-embedded. The organizations that thrive will be those that elevate what their teams can achieve, ground every initiative in measurable results, move forward with confidence and responsibility, and build capabilities designed for what’s next.
We help organizations navigate this path with practical, outcome-focused guidance. We meet clients where they are in their AI journey, bring the lessons we have learned inside our own firm to bear on theirs, and partner with them to design and enable their AI program from ambitions to trusted outcomes - impact that lasts beyond any single technology cycle.
Connect with our AI leaders for a strategic conversation about your priorities, current maturity, and the practical next steps you can take this quarter.