Summary Notes and Webinar Recording: Smarter with AI – Productivity Without Losing Judgement.

On 13 October 2026, Qadoos Rashid, former Group Chief People Officer at European Dental Group, and Johannes Sundlo, Founder of Prorio AI and CRF’s AI in HR Course Director, joined CRF for the second webinar in its AI in HR Series on Smarter with AI – Productivity Without Losing Judgement.

AI can help people complete tasks faster, but speed does not necessarily translate into better performance. Used poorly, it can encourage over-reliance on convincing outputs, weaken professional judgement and accelerate work that creates little value. Used effectively, it can free up capacity, improve decisions and enable people to focus on work that matters.

Building on the previous webinar, Where Does AI Really Create Value?, this session explored how organisations can improve productivity without outsourcing human judgement. Drawing on practical examples from European Dental Group, the discussion considered where AI should accelerate work, where it should challenge professional thinking and what organisations must do to turn technological capability into measurable business outcomes.

View the summary notes and key takeaways or watch the webinar recording below:

Understanding Judgement in an AI-enabled Workplace

As AI becomes embedded in everyday work, organisations need to reconsider what good professional judgement looks like.

Johannes Sundlo compared this shift with the evolution of farming. Modern farmers rely on technology, machinery and weather forecasts to make decisions that previous generations approached differently. Their expertise has not disappeared, but the knowledge and capabilities required to perform their work have changed.

AI may bring about a similar transition. The question is not simply whether employees are relying too heavily on technology, but which aspects of judgement should remain human, how people should evaluate AI outputs and who is ultimately accountable for decisions.

Human experience remains essential. If organisations rely exclusively on AI-generated recommendations, they risk producing increasingly similar approaches and decisions. Combining technological capabilities with professional knowledge, contextual understanding and individual perspectives creates opportunities for better outcomes.

However, this cannot be left to chance. HR and business leaders must make expectations explicit, helping employees understand when to accept, challenge, revise or reject AI-generated outputs.

Three practical applications: AI as an assistant, challenger and mirror

Qadoos Rashid shared three examples from European Dental Group (EDG), illustrating how AI can play different roles in improving workforce productivity while preserving professional accountability.

1. Nova: Releasing valuable clinical capacity

In dentistry, recruitment challenges are often seen as the primary constraint on business performance. However, increasing headcount does not necessarily increase productive capacity.

Analysis of EDG’s Dutch business identified substantial differences in revenue per clinical hour across its workforce. Dentists under 30 generated approximately €250 per clinical hour, compared with €300 among those aged 36 to 45, a gap of around 17%. Meanwhile, the proportion of clinical days delivered by the younger cohort had increased from approximately 6% to 19%.

This was not evidence that one generation was inherently more productive than another. Instead, it highlighted the need to adapt development, support and ways of working as the workforce changed. EDG estimated an opportunity worth approximately €5 million in annual revenue, or €2 million in EBITDA, in the Netherlands alone.

One important source of lost productivity was unused appointment capacity. When a patient cancelled at short notice, reception teams had limited time to contact suitable replacements. If the appointment remained empty, valuable clinical capacity was permanently lost.

EDG developed Nova, an AI-powered patient access assistant that identifies suitable patients, matches their treatment needs with available appointments and contacts them automatically.

The objective was not to replace receptionists. It was to recover clinical capacity while allowing overstretched reception teams to devote more attention to patients requiring personal support.

In the Netherlands, Nova delivered more than €3 million in run-rate EBITDA. Patients benefited from earlier access to treatment, clinicians made better use of their available time and reception teams were released from repetitive administrative work.

The lesson is that AI creates value not simply by completing tasks faster, but by enabling the organisation to use its existing capacity more effectively.

2. Pearl: Improving decisions through constructive challenge

EDG’s second example, Pearl AI, demonstrates a different role for technology.

Pearl analyses dental X-rays and highlights potential areas of concern for clinicians to investigate. Rather than making the final decision, it provides another perspective that professionals can use to challenge their own assessments.

The technology was introduced across EDG’s UK business, where the challenge was less about acquiring the tool and more about embedding it into clinical practice.

Qadoos described his own experience as a patient, when two experienced dentists and Pearl reviewed his dental X-rays. The three perspectives did not always agree. Clinicians questioned the AI’s findings, considered additional patient information and discussed which issues required treatment or further investigation.

Some AI-generated recommendations were accepted, others revised and some rejected. In one instance, the technology prompted clinicians to reconsider an issue they had initially evaluated differently.

This illustrates the importance of maintaining the competence, authority and accountability to disagree with AI. Human oversight is meaningful only when professionals can independently assess the evidence and exercise judgement.

Pearl also helped clinicians communicate more effectively with patients by making potential problems easier to see and discuss. EDG observed stronger treatment plan adoption where the technology was embedded. Pearl contributed to wider operating improvements during a period of double-digit growth in the UK business, although it was not presented as the sole cause of that growth.

Unlike Nova, which demonstrated where AI could accelerate work, Pearl showed how technology can create value by encouraging professionals to slow down, reconsider the evidence and make better-informed decisions.

3. AI Copilot: Supporting reflection and professional development

The third example involved an early-stage pilot exploring whether an AI copilot could support dentists in understanding and improving their own productivity.

Providing meaningful feedback to experienced professionals can be difficult, particularly where individuals value their clinical autonomy and may be reluctant to accept recommendations from those without equivalent expertise.

EDG began exploring whether a confidential AI copilot could provide a more consistent and psychologically safe environment for reflection.

Using carefully governed workforce and performance data, the proposed tool would help clinicians identify patterns in their work, explore potential development opportunities and consider where improvements might be possible.

Crucially, the copilot would not prescribe treatments, make performance judgements or replace human coaching. Clinicians would remain responsible for interpreting the information and deciding what action, if any, to take.

Qadoos described this as using AI as a mirror rather than a manager.

The initiative remains in the pilot stage, with no outcomes yet established. Its potential lies in supporting individual reflection and professional development without undermining autonomy or accountability.

Five key takeaways

The three examples illustrate that AI has no single role in improving productivity. Different applications require different levels of automation, oversight and human involvement. Five lessons emerged from the wider discussion.

1. Start with the business problem, not the technology

Organisations should begin by identifying the outcomes they need to improve rather than looking for opportunities to deploy AI.

In EDG’s case, the starting point was not how to automate reception work, but how to address lost clinical capacity, improve patient access and make better use of an increasingly scarce workforce.

This distinction matters. AI can make an existing process faster without resolving the underlying problem. A successful intervention requires a clear understanding of the business challenge, its causes and the outcomes that would constitute meaningful improvement.

HR can contribute by connecting workforce data with operational performance, challenging assumptions about headcount and helping the business identify opportunities that might otherwise be overlooked.

2. Productivity gains depend on how released capacity is used

Time saved is not inherently valuable. Its contribution depends on what employees do with the capacity AI makes available.

Nova illustrated this principle. Automating patient contact reduced repetitive work, but its broader value came from filling unused clinical appointments and allowing reception teams to spend more time supporting patients.

This requires organisations to plan ahead. Before introducing AI, leaders should consider which activities will change, what employees will stop doing and where their time and attention can create greater value.

Without these decisions, organisations risk achieving isolated efficiency gains without improving overall performance.

3. Human judgement must be actively developed, not simply protected

The discussion highlighted concerns that widespread AI use could weaken the capabilities employees need to assess information, solve problems and make independent decisions.

Johannes pointed to the potential implications for early-career professionals. If AI takes over tasks through which people traditionally gain experience, organisations may need new approaches to developing expertise.

He described a conversation with a university professor who had observed that students were producing increasingly polished academic work, but with less evidence of their own perspectives and independent thinking.

Similar challenges may emerge in the workplace. Employees could become highly effective at producing outputs while losing opportunities to develop the knowledge and experience required to evaluate them.

HR must therefore consider not only which tasks AI can perform, but which activities contribute to learning, professional development and the formation of judgement.

This does not mean preserving every existing task. It means being intentional about which capabilities remain important and designing opportunities for employees to develop them.

4. Knowing when to accelerate and when to slow down is critical

Not every task requires the same degree of human involvement.

High-volume, repetitive and relatively rules-based activities may be appropriate for automation. More sensitive decisions, particularly those affecting people, professional standards or organisational values, require greater scrutiny.

Johannes offered difficult management conversations as an example. Although AI could technically conduct such conversations, organisations must consider whether delegating them would be consistent with how they want to treat employees and develop their leaders.

Similarly, Pearl demonstrated that AI-generated findings should sometimes prompt additional investigation rather than an immediate decision.

The challenge is to distinguish between tasks where speed creates value and those where careful consideration, relationships and professional judgement are essential.

5. Successful AI adoption requires leadership, collaboration and clear accountability

Introducing technology is only part of the challenge. Organisations must also create the conditions for people to use it effectively.

At EDG, AI initiatives involved collaboration between HR, finance, operations and technology leaders. Senior leaders tested tools themselves before introducing them more widely, while respected clinical experts helped demonstrate their value to colleagues.

Each pilot required a defined problem, clear ownership, appropriate safeguards, a baseline and criteria for deciding whether to scale, adapt or stop.

Johannes reinforced the importance of integrating AI into everyday work rather than treating it as a separate technology initiative. As adoption grows, leaders need to understand how AI changes performance expectations, workforce planning, capability requirements and the management of teams.

Building technical familiarity is necessary, but insufficient. Organisations must also develop the leadership capabilities, learning environments and management practices needed to realise AI’s potential.

Q&A Corner

The discussion explored several practical questions about implementing AI, measuring its impact and maintaining professional judgement.

1. How should organisations measure whether AI is creating value?

Measurement should reflect the original business problem rather than the capabilities of the technology.

For Nova, EDG examined the commercial impact of recovering unused clinical capacity alongside improvements in patient access and the work undertaken by reception teams.

For Pearl, attention focused on how the technology supported clinical discussions, patient understanding and treatment plan adoption.

These measures extend beyond efficiency. They consider whether AI improves outcomes for employees, customers and the organisation.

For initiatives still being tested, such as the clinical productivity copilot, success measures must be established before wider deployment. Where the evidence does not demonstrate value, organisations should be prepared to adapt or discontinue the intervention.

2. How long does it take to implement and embed AI successfully?

The timeframe depends on the nature of the intervention.

Qadoos explained that Nova was developed and implemented within approximately six months. Pearl, by contrast, was an existing technology, but required around a year to become embedded in clinical practice.

This highlights the difference between making technology available and changing how people work.

Although AI can accelerate software development, successful adoption often depends on building trust, changing professional behaviours and helping people understand how to apply the technology appropriately.

3. How can organisations build employee trust and encourage adoption?

Credible leadership and practical demonstration are important.

At EDG, respected clinical leaders helped build confidence by showing how they used AI within their own work and how it supported better outcomes.

Senior business leaders also experimented with AI tools before extending access, ensuring that they understood both the potential benefits and limitations.

Johannes emphasised that organisations must move beyond simply training employees to use individual tools. Managers also need support in setting expectations, evaluating AI-assisted work and creating environments where employees can question outputs and learn from experimentation.

Trust develops when people understand why a tool is being introduced, how it will affect their work and where responsibility for decisions remains.

4. What should organisations prioritise when deciding where to invest in AI?

Qadoos encouraged organisations to focus on opportunities that could make the greatest difference to business performance rather than limiting their ambitions to small, easily automated tasks.

At EDG, this thinking extended beyond individual applications to the possibility of redesigning AI-enabled practice management systems across Europe, potentially affecting interactions with patients, employees and suppliers.

Johannes emphasised the need to make deliberate choices about which activities should remain human and where AI can contribute more effectively.

Both perspectives point to the same conclusion: organisations should not adopt AI simply because it is available. They should concentrate on the opportunities that create meaningful value and be explicit about the role of human judgement.