A leadership team asks HR what skills the organization will need in five years. Someone suggests prompt engineering. Another proposes hiring a data scientist. A third wants a quantum readiness programme because a competitor has mentioned quantum computing in a presentation.

The meeting produces a training wish list, but no agreement about which business problem the organization needs to solve. That’s where future skills planning often loses its usefulness. The vocabulary moves ahead of the work.

Canadian employers need a more disciplined approach. Start with the processes that matter to customers, costs and risk. Identify how those processes could change, determine what people must be able to do and fund the capabilities that the evidence supports. Longer-term technologies belong in the plan, with explicit conditions for increasing investment.

HR doesn’t need a precise forecast of future job titles to lead that process. It needs a practical method for making workforce decisions while technology, business demand and the available evidence keep changing.

Use Canadian evidence without turning it into a prediction

In the third quarter of 2026, Statistics Canada found that 25.2% of businesses planned to use AI to produce goods or deliver services over the following 12 months. Planned use varied by industry, reaching 54.8% in professional, scientific and technical services, compared with 8.0% in transportation and warehousing. [1]

Those differences matter. An employer shouldn’t adopt the staffing assumptions of a professional services firm simply because both organizations operate in Canada. Its exposure depends on its processes, customers, technology and capacity to implement change.

The survey measures intentions. It doesn’t establish which projects will succeed, how much productivity will improve or which employees will be displaced. HR can use the findings to inform questions about its own business, while resisting the temptation to convert a national adoption figure into a local headcount target.

The same discipline applies to technology forecasts. A potential application is a reason to investigate. It becomes a workforce investment when the organization can explain the business need, the capability gap and the evidence required to proceed.

Separate the technologies before setting the budget

AI, automation and quantum computing are often placed in the same sentence. They create different implementation demands, and HR should understand enough about each to ask useful questions.

Automation uses technology to carry out work with less manual intervention. A rules-based invoice workflow, a conveyor system and an industrial robot can all change jobs without using generative AI. Workforce needs may include process configuration, maintenance, troubleshooting, safe operation and managing exceptions.

AI covers a broad range of systems that perform tasks such as recognizing patterns, making predictions or generating content. Generative AI can produce text, images and other outputs. The people using it need to understand the work well enough to check whether those outputs are suitable. A credible-looking response can still be inaccurate or unsupported.

Quantum computing uses principles of quantum physics to process information in ways that may help address certain classes of problems. It isn’t a general replacement for ordinary computers or a faster version of a workplace chatbot. Commercial usefulness depends on the application and technical maturity.

Canada’s quantum computing roadmap identifies the development of a fault-tolerant quantum computer as a long-term goal and describes limitations involving errors and qubit counts. That supports targeted research and monitoring rather than an assumption that every employer needs a broad quantum retraining programme immediately. [2]

For HR, the practical distinction is budgetary. A deployed automated production line may require immediate operator and maintenance training. An AI document workflow may justify a supervised pilot. A speculative quantum optimization use case may justify assigning someone to monitor developments and maintain contact with a research partner.

Map the work before discussing new roles

Choose one process with a measurable business consequence. Examples include resolving customer requests, preparing estimates, scheduling maintenance or reviewing supplier invoices. Avoid beginning with an organization-wide inventory that takes months to assemble and produces little that managers can use.

Walk through the process with the people who perform it. Record the inputs, decisions, handoffs, outputs and common exceptions. Identify where employees wait for information, repeat work or correct errors. These details often explain more about the opportunity than the job description does.

Next, identify the proposed technology change. Will the system retrieve information, recommend a response, execute a transaction or make a decision? Each creates a different requirement for human involvement. A draft that an employee reviews has a different risk profile from a system that sends a contractual commitment directly to a customer.

For each changed activity, specify the remaining human responsibility. Someone may need to verify the source data, approve an exception, explain the result or stop the workflow. Name that responsibility before calculating the staffing effect. Otherwise, essential work can disappear from the plan while remaining necessary in practice.

The resulting map should also identify dependencies. An employee can’t reliably check a recommendation without access to the relevant record. A supervisor can’t approve exceptions promptly if the implementation centralizes every unusual case in an already overloaded role.

Describe capability in terms managers can assess

A skills list that includes adaptability, digital literacy and critical thinking sounds reasonable but provides little guidance for selection or training. Translate those terms into observable work.

For an employee using AI to draft customer correspondence, competence might mean checking the customer’s actual agreement, identifying an unsupported claim, protecting personal information and escalating a proposed concession beyond their authority. For a production supervisor, it might mean recognizing an abnormal machine condition and following the approved shutdown and escalation procedure.

A useful capability statement identifies the task, the conditions, the acceptable standard and the limit of authority. For example, an employee may approve a routine invoice within a defined limit after matching specified records, while escalating changed bank details through an independent verification process.

That description helps HR distinguish between a training gap and a system problem. If employees can’t perform because information is inaccessible or approval rules conflict, another course won’t solve the problem. Process owners need to fix the conditions under which the work occurs.

It also creates a fairer basis for evaluating candidates. A relevant work sample can reveal transferable capability that a job title or a list of software keywords misses. Use consistent scoring and provide appropriate assessment conditions so the exercise measures the intended ability.

Develop the skills around the technology

Some employers will need technical specialists. Many will also need people who can translate operational needs into workable changes and maintain the resulting process.

A knowledgeable supervisor may be well placed to identify meaningful exceptions in a production workflow. A customer service employee may understand why a standard answer fails for certain accounts. A procurement employee may recognize the difference between a supplier’s impressive demonstration and a service commitment the employer can enforce.

HR should look for that existing knowledge before assuming every capability must be recruited externally. Pair employees who understand the work with technical expertise, and give both groups time to learn from each other. The operational employee should be able to explain the business consequence of an error; the technical colleague should be able to explain the system’s limits.

Management skills also change. Supervisors need to allocate review work, interpret new performance measures and respond to employees who challenge an automated output. If managers reward speed while treating escalations as poor performance, the organization may undermine the very judgment it claims to value.

There’s a retention issue as well. Employees who contribute to a successful implementation may take on more demanding responsibilities. Review their job scope, development opportunities and compensation rather than assuming the change is simply another task within the existing role.

Give quantum a proportionate place in the plan

For many employers, the first useful quantum discussion will involve information security. A sufficiently capable future quantum computer could threaten some widely used public-key cryptography. Preparing for that risk involves understanding systems, data and suppliers, along with the skills needed to manage a transition.

The Canadian Centre for Cyber Security’s roadmap for the Government of Canada sets milestones for migrating non-classified federal systems to post-quantum cryptography. These include completion for high-priority systems by the end of 2031 and remaining systems by the end of 2035. Those are federal roadmap milestones, not blanket statutory deadlines for every Canadian private employer. [3]

Post-quantum cryptography refers to cryptographic methods designed to withstand attacks from quantum computers and can run on conventional systems. An employer doesn’t need to buy a quantum computer to begin preparing its security architecture.

HR’s role is to ask its technology and security leaders what capability the organization needs to manage this work. Who understands the relevant systems? Who can obtain credible supplier commitments, manage migration projects and retain the knowledge internally? If that expertise sits entirely with one contractor, how will the employer maintain informed oversight?

Organizations closer to quantum research, specialized manufacturing or relevant commercial applications may need a more substantial talent strategy. That can involve specialist recruitment, education partnerships and longer development periods. The appropriate scale should follow a defined business or research programme, with technical leaders validating the skills required.

For other employers, a named monitoring owner and a scheduled review may be sufficient for now. Write down what would change that decision, such as a customer requirement, a supplier migration programme or a validated application with a credible business case.

Compare hiring development and external support honestly

Consider a hypothetical Canadian business introducing an AI-assisted document review workflow. It needs someone to own the process, coordinate implementation and monitor quality. Management is comparing an external hire, internal development and a specialist partner.

The figures below are illustrative Canadian-dollar assumptions. They aren’t salary benchmarks or supplier quotations. Each option is intended to address the same initial operating need, but management must test whether the proposed capacity and competence are genuinely comparable.

An external hire might involve $125,000 in annual loaded employment cost, $15,000 in recruitment expenses and $10,000 in onboarding and integration costs. That produces a first-year planning total of $150,000. The organization still needs to consider recruitment lead time and who covers the work before the person starts.

An internal development option might allocate half of an existing employee’s capacity, valued at $45,000, plus $20,000 in training and coaching, $30,000 in backfill and $20,000 in external technical support. The total is $115,000. The allocated employee time is an opportunity cost; it isn’t an additional salary payment. Finance should check that the backfill doesn’t duplicate a cost already included elsewhere.

A partner option might involve $90,000 in annual fees, $20,000 in internal coordination capacity and $15,000 in knowledge transfer and exit preparation, for a total of $125,000. Its suitability depends on service scope, availability, data arrangements and whether the employer retains enough expertise to direct and assess the work.

The cheapest total isn’t automatically the best decision. Internal development may take longer than the project allows. A specialist hire may provide capacity well beyond the initial need. A partner may accelerate implementation but create an ongoing dependency. Compare the expected time to competence, quality, continuity and recurring cost, then record the assumptions that could make the preferred option fail.

Finally, value the benefit separately. If the project releases 1,000 hours at an assumed $45 an hour, that represents $45,000 of capacity. It becomes a cash saving only if spending falls. If the hours support more work, the business needs a credible path to additional output or revenue. Counting both the full labour saving and the revenue from the same released hours would overstate the benefit.

Put employment and privacy review inside the project

Technology adoption can change responsibilities, workloads and the evidence used to assess employees. HR should review those consequences before implementation, alongside the applicable employment, labour, human rights, privacy and safety requirements.

The legal position depends on jurisdiction, sector, contracts, collective agreements and the proposed change. A generic technology policy won’t determine whether a particular restructuring, monitoring practice or selection process is appropriate. Obtain advice on the actual design where needed, while the team still has room to change it.

Canada’s privacy regulators emphasize that organizations developing or using generative AI remain responsible for compliance with applicable privacy laws. [4] For workforce planning, this means data handling competence must be part of the role design. Employees need approved tools, clear permissions and a workable process for reporting mistakes.

Consult the people affected by the change. Ask how it alters their work, what makes a recommendation difficult to verify and which targets could encourage shortcuts. Explain what management has decided, what remains under review and how redeployment or development opportunities will be assessed. Don’t promise employment security that the organization hasn’t authorized.

Build a useful plan in 90 days

A focused 90-day exercise can give executives a credible investment decision without pretending to solve the organization’s entire future. Select one process, appoint an operational sponsor and involve HR, finance and the appropriate technology, security and legal advisers.

During the first 30 days, establish the baseline. Measure workload, turnaround time, rework, supervision and significant exceptions. Map the proposed change and identify the capability gaps. Include the employees who understand how the process behaves on a difficult day, rather than relying solely on its documented procedure.

By day 30, the team should be able to explain the business problem in plain language. It should also know which result would justify proceeding. A statement such as improving response time while maintaining an agreed quality threshold is more useful than an ambition to become AI ready.

During days 31 to 60, run a bounded pilot with approved data, clear responsibilities and an escalation route. Protect time for practice and review. Include realistic exceptions rather than demonstrating the system only on straightforward cases. Compare outcomes with the baseline and record the review effort required.

By day 60, the team should have evidence about whether the tool helps and whether employees can operate the changed process reliably. Distinguish a technical failure from a skills gap. If the system requires impractical levels of checking, training employees harder may simply conceal a poor design.

During days 61 to 90, decide whether to expand, revise or stop. Compare the workforce options, confirm the cost of achieving competence and review the employment implications. Establish who owns the process after the project team disbands. Include a development pathway for backup coverage and a plan for transferring knowledge from suppliers.

The executive submission should identify the expected benefit, workforce investment, main dependencies and conditions for review. A smaller employer may need only a short decision paper. The value comes from the evidence and ownership, rather than the size of the document.

Use triggers to keep the plan current

A workforce plan becomes stale when its assumptions remain unchanged despite new evidence. Schedule reviews, but also identify events that require an earlier decision.

A material increase in error rates may trigger additional review or a pause. A recurring customer requirement may justify developing a specialist capability. A supplier’s withdrawal of support may require a different sourcing model. A successful pilot with manageable oversight may justify extending training to another team.

Set thresholds with the relevant process owner. They should reflect the consequence of failure, available data and tolerance for risk. A low-consequence drafting task and a safety-critical operational decision require different standards. Avoid presenting one universal accuracy percentage as proof that all applications are ready.

Keep track of what employees can now do, how much review they need and whether the organization has backup coverage. Course completion and licence usage can help explain participation, but they don’t establish that the business has acquired the capability it needs.

At the next planning cycle, HR should be able to say which investment improved performance, what competence it created and what evidence supports the next decision. That’s a stronger contribution than naming the job titles that might become fashionable. It gives the employer a workforce strategy that can change with the work and a clear basis for deciding where the next dollar belongs.