K3 Insights

Welcome to the K3 hub

9 min read - November 16, 2026

Creating Greater Support Function Efficiencies in an AI-Enabled World

I have previously outlined my perspective on why HR as a profession needs to challenge itself, becoming far more of a strategic enabler and operational efficiency engine through the use of AI.

This will allow HR to become more of a core architect of how the organisation performs, not only because it is operating in the right way, but because it is seen as an exemplar of what good looks like in an AI-enabled business environment.

And the potential opportunity is significant.

Gartner estimates that 50% of HR work could be automated or agent-handled by 2030 [1]. It also reports that 81% of HR leaders investing in AI already see a positive impact on productivity, with 31% agreeing this investment has reduced overall HR costs [2].

But this isn't just an HR issue.

Research across other enabling functions points in a similar direction:

  • Thomson Reuters reports that 77% of legal professionals expect agentic AI to be central to legal workflows by 2030. It also cites evidence that 65% of attorneys using AI already save 1–5 hours per week, with some saving 6–10+ hours [3].
  • McKinsey estimates agentic AI could power as much as two-thirds of current marketing activities, with approximately 60% of tasks across the marketing process potentially agent-powered [4].
  • Gartner predicts that by 2030 agentic AI will make at least 15% of day-to-day finance decisions autonomously, while 70% of finance functions will use AI for real-time decision-making [5].

For me, this research suggests an opportunity to fundamentally redesign how work is performed within these functions, not just automate some isolated tasks, while also acknowledging that the scale of opportunity will vary by function and workflow.

The problem isn't necessarily seeing the opportunity

As a specialist in operating model design, over the last 18 months I have increasingly been working with organisations looking to drive efficiencies into their enabling functions, so that customer-facing teams can get access to more resource - whether that is time, money or people.

The issue I see is that many organisations don't know how to review their workflows through an AI lens or, somewhat ironically, they are simply too busy to free up the capacity to do the work.

Technology is another constraint.

Many organisations have platforms underpinning these functions - from SuccessFactors and Workday through to BambooHR and HiBob in HR, for example - and increasingly these platforms provide opportunities to overlay AI onto existing work.

But many organisations are also suffering from years of underinvestment in enabling-function technology. One large utility I have worked with has a 20-year-old ERP system that Finance must use for all its critical work.

McKinsey has previously highlighted how enabling functions have often been managed primarily as a cost base rather than as a source of technology-enabled value [6].

The result is that organisations can see the potential of AI, but struggle to translate it into meaningful changes in how work actually gets done.

Start with the work, not the technology

My proposition is relatively simple.

Enabling functions have reasonably standard lifecycles and activities. These can be mapped, redesigned through the use of AI and, with the support of Technology teams or vendors, potentially overlaid across existing systems.

There is often significant commonality across core HR or Finance processes. So rather than every organisation starting from scratch, much of the approach could be developed as a generic starting point and then customised.

That could speed up the work, remove some of the barriers to getting started and, in some circumstances, reduce or defer the need for costly technology replacement.

The starting point is to understand the work. For each function, this means mapping the key activities across its lifecycle, comparing the current and desired future state, and identifying the priorities for change that will better support delivery of the organisation’s strategy.

Once those priorities are understood, the next step is to map the underlying workflows — breaking them down into the key value points and activities. This creates the foundation for assessing where AI could be applied and, importantly, where it could create the greatest value.

Recruitment provides a useful example

Once a priority workflow has been mapped, the next question is not simply “Where can we use AI?”, but “What is the appropriate role of AI in each activity?”

One tool I have found useful for this is Slalom's AI Integration Spectrum [7], which considers different levels of AI involvement — from relatively simple assistance (red) and augmentation (orange) through to automation (purple) and agentic activity (pink).

Applying this across the Recruitment workflow highlights both the scale of the opportunity and the different roles AI could play.

As the example demonstrates, there is considerable opportunity for AI disruption across the workflow. But importantly, some activities remain deliberately human-led.

That distinction matters. The objective shouldn't be to automate everything simply because we can. It should be to determine where technology genuinely creates value and where human judgement, interaction or accountability remains important.

From AI opportunity to impact

Identifying where AI could be applied is only part of the equation. The next step is to understand what difference those changes could actually make to the people doing the work.

In the example below, I have assessed the potential impact across the first three stages of the Recruitment workflow for both the Hiring Manager and Recruiter.

The potential gains are significant, but they are not evenly distributed. Some interventions primarily release recruiter capacity, while others significantly reduce the effort required from Hiring Managers. Some activities could be largely automated, while others continue to require meaningful human involvement.

This is an important distinction because the opportunity isn't simply to implement AI or reduce effort. It is to redesign the work, release capacity and determine how that capacity can be redirected to higher-value activities.

That is where the conversation starts shifting from AI implementation to operating model redesign.

Finding the efficiency is only half the job

This is also where many organisations appear to get stuck.

Gartner reports that only 5% of HR functions have integrated AI into their operating models and only 18% report significant value, with a third reporting no or minimal value [1].

Based on my experience to date, part of the issue is that teams can identify problems in the current state and even identify potential AI opportunities, but are less clear about what needs to change around the technology to actually realise the gains.

For our Recruitment example, that means looking beyond technology.

  • Process and systems need to change. The end-to-end workflow needs to be redesigned, integration with the broader technology ecosystem considered, and the efficiency gains quantified.
  • Structure and governance also need to change. As transactional activity reduces, roles and capacity need to be redesigned to focus more heavily on the "human moments that matter". Decision rights also need to be explicit about where human approvals remain mandatory.
  • People, culture and leadership need attention. Hiring Managers need to understand how their role is changing and how to use AI appropriately. Recruiters need the capability to operate effectively in the new model and understand where human-first interactions remain critical.

There are also very real risks to manage - bias, hallucinations, privacy and data security, transparency and explainability, and candidate experience among them.

There is an important organisational values question here too. If AI can take 100+ applications and produce a shortlist without a human ever reviewing the applications, does the organisation want it to? The fact that something can be automated doesn't automatically mean that it should be.

Prove the value, then scale it

From the outset, I would advocate defining a set of "current to future state" measures.

For Recruitment these might include time to hire, recruiter hours per hire, cost, candidate experience, and identified AI-related errors or exceptions.

That establishes the baseline, helps justify the investment and, importantly, provides a way of determining whether the redesigned workflow is actually delivering what was promised.

And then start small.

Showcase value. Embed the change. Learn from it. Then pick off the next priority.

Because this isn't "one and done". Technology will continue to evolve, as will the organisation's ability to use it.

The enabling functions also shouldn't be "doing this change" to the business. In Recruitment, for example, recruiters should be working with Hiring Managers to understand requirements, shape the options and pilot the changes.

That creates greater ownership of the process change - and reinforces that the purpose is ultimately to help people do their day job better, not simply give them another transformation programme to participate in.

The change process also needs to factor in diversity and ethical considerations. Does the evolved AI-enabled workflow have detrimental socioeconomic, gender or ethnic impacts? Are we comfortable with the decisions we are delegating to technology? And are those decisions consistent with the organisation's values?

The bigger opportunity

Organisations are operating in rapidly evolving marketplaces and, if they are going to remain competitive, it isn't just the customer-facing teams that need to change.

Competitive advantage will increasingly come down to how effectively the work of enabling functions is redesigned: freeing up capacity through AI, ensuring the right roles are focused on the right work and keeping these functions relevant.

The opportunity I see is therefore to make this a repeatable approach.

Start with a typical enabling-function workflow. Map the work. Identify the appropriate level of AI involvement. Quantify the opportunity. Redesign the operating model elements required to capture it. Measure the results. Then scale what works.

There is no shortage of discussion about what AI might do to organisations.

The more important question now is how we actually redesign the work to capture that value.

References
[1] Gartner (2026) – Stop Finding Features, Start Building Value: How CHROs Win with HR AI

[2] Gartner (2026) – 3 CHRO Actions to Drive Measurable AI Impacts

[3] Thomson Reuters (2026) – What Legal Professionals Need to Know About Agentic AI

[4] McKinsey (2026) – Reinventing Marketing Workflows with Agentic AI

[5] Gartner (2026) – Finance 2030: 8 Forces Shaping the Future of Finance

[6] McKinsey (2022) – How Can Corporate Functions Become More Agile?

[7] Slalom (2026) – AI Integration Spectrum

Back to Articles

Contact us