Views: 28

What the Workforce Numbers Tell Us

Boards need to know where AI could change work, how much capacity it could release, and which skills the organisation will need next. The World Economic Forum’s Future of Jobs Report 2025 provides a starting point. It draws on responses from more than 1,000 employers across 55 economies.

  • 39%: The share of workers’ existing skill sets that employers expect to change or become outdated by 2030.
  • 63%: The share of surveyed employers that identified skills gaps as a barrier to business transformation.
  • 41%: The share of surveyed employers that expected to reduce their workforce where AI can automate tasks. That figure describes employer intentions, with no estimate of the share of jobs that will disappear.

For a board decision, translate those broad trends into local questions. Which tasks are changing? Who performs them today? Where could freed capacity be used, and what training would make that possible?

Why These Sectors Are Worth Examining

Banking, insurance and professional services all have workflows built around collecting documents, evaluating evidence and making decisions. These workflows provide concrete places to test AI. The workforce effect depends on how much of the work changes, which checks still need people, and whether demand rises as turnaround improves.

Banking

McKinsey’s analysis of agentic AI in banking operations describes changes to end-to-end workflows. Start with tasks your bank can measure, such as loan documentation, compliance reporting, customer service records or reconciliation. Separate routine cases from exceptions and record the time each takes.

In its 2023 economic potential study, McKinsey estimated that generative AI could create US$200 billion to US$340 billion in annual economic value for banking globally if the use cases it analysed were fully implemented. It is a sector-level estimate of potential value. A bank’s own business case needs its workload, implementation costs and controls.

Model the effect by role and express it in full-time-equivalent (FTE) capacity. Then decide whether the time released could improve service, absorb higher volumes or support redeployment. Use the estimate to identify which employees need training and which teams could use their skills.

Governance and the EU AI Act

The European Commission’s AI Act overview explains that certain AI uses in employment, such as CV sorting for recruitment, can be high-risk. A workforce model that estimates team capacity raises different questions from a system used to screen or evaluate individual employees. Identify the intended use and check with Legal which obligations apply.

The Commission’s enforcement timeline lists 2 December 2027 for high-risk systems covered by Annex III. For workforce planning, document assumptions, data sources and approval decisions so the board can trace the model and understand where human review takes place.

Insurance

Oliver Wyman’s property and casualty analysis estimates 5% to 20% potential time savings in claims workflows. The estimate varies by the type of claim, line of business and share of documentation that can be automated. Test local handling times and include the time adjusters spend checking AI outputs and resolving exceptions.

Underwriting illustrates another outcome. Accenture reports that QBE could process 100% of broker submissions in the product lines where its AI-assisted solutions were in production. That case measures coverage in specific product lines. An insurer planning its own pilot should track coverage, turnaround time and the quality of risk decisions separately.

EIOPA reports that 65% of 347 surveyed insurers across 25 European Economic Area countries were already using generative AI, and 23% planned to use it. The survey measures adoption. To assess workforce readiness, ask which teams have measured the time saved and agreed where to use it.

Accounting and Professional Services

ACCA’s work on skills for responsible AI adoption provides a starting point for assessing what staff will need to use and review these tools. Examine tasks such as preparing reconciliations, summarising records and drafting client documents. Record which outputs require verification, professional judgement or sign-off.

If AI changes the work junior staff usually perform, examine what those tasks teach. An associate who spends less time preparing a reconciliation still needs to learn how to investigate a discrepancy. Plan supervised practice, review and client exposure alongside changes to staffing and throughput.

Questions the Board Will Ask

  • Which tasks and roles are affected, and how many FTE hours do they represent?
  • Where could capacity increase or decrease over the next three years under different rollout scenarios?
  • How much of the potential benefit is released time, additional throughput or a change in expenditure?
  • Which skills must teams develop, and where could employees move?
  • Which AI uses need legal review, oversight or additional documentation?

Report the assumptions behind each answer and identify which figures come from observed results. That gives the board a basis for asking what would change if adoption were slower or oversight took longer than planned.

What Each Planning Input Can Contribute

1. Industry Benchmarks

Use benchmarks to identify questions for a pilot. Estimate the effect on a specific workflow from your own volumes, handling times, staffing and service requirements.

2. HR Information Systems

Use role, headcount, payroll and skills data as a baseline. Connect that information to workflow volumes and the time staff spend carrying out and checking each task.

3. External Advice

Consultants can provide specialist input or challenge assumptions. Give HR and Operations ownership of the model so they can update it when tools, volumes or staffing change.

What a Board-Ready Workforce Model Includes

  • Task volumes, handling times and exception rates by function.
  • The share of work AI can support, adjustments for adoption, and time needed for human review.
  • Released FTE capacity by role, with redeployment, training and staffing scenarios.
  • One-time and ongoing costs, including integration, licensing and oversight.
  • Named owners, review dates and a plan to check projected outcomes against pilot results.

Illustrative example: If a claims team handles 2,000 eligible cases per month and a pilot saves a net six minutes per case after review, that releases about 200 staff hours per month. At 150 productive hours per FTE per month, that is about 1.3 FTE of capacity. Finance can then assess whether the time would absorb more claims, reduce overtime or change spending.

When to Start Planning

Start while the pilot still allows you to collect a reliable baseline. Schedule legal and risk reviews for the specific AI use, give teams time to learn new workflows, and decide what evidence you need before expanding the rollout.

Plan for Released Capacity and Skills Shortages Together

A workflow can require fewer hours of document preparation and more time for exception handling, quality checks or customer conversations. Model both changes. Compare the skills available in the teams likely to release capacity with those needed in teams where workload is rising.

Where to Start

1. Map the Work

Choose one function and list the tasks, volumes, handling times and judgement required. Record who owns each workflow and who can validate the baseline.

2. Translate the Impact Into Capacity and Cost

Model realistic adoption and subtract the time spent checking outputs and handling errors. Compare the impact across roles and identify training and redeployment options.

3. Bring a Testable Proposal to the Board

Present the expected benefit, the conditions it depends on, and the decision the board needs to make. Set a review date and name who will compare actual outcomes with the model.

EDLIGO AI Workforce Transformation

EDLIGO AI Workforce Transformation helps teams map AI exposure at task level and translate it into estimated FTE capacity by role and function using their own data.

Book a 30-minute meeting to discuss a diagnostic for banking, insurance or professional services. Bring a workflow you want to evaluate so you can review the data needed, the capacity scenarios and potential skills implications.

What to Bring to the Next Board Meeting

Bring a baseline, one measured pilot result and a breakdown of where any time saved is expected to go. Show what the team has validated, what remains an assumption, and when you will revisit the workforce and investment decision.

WordPress Cookie Plugin by Real Cookie Banner