Governance Is Not the Brakes: Rethinking AI Delivery at Scale

Jan Esman

Jeroen Bolluijt

This article draws on insights from Inside AI: Governance Is Not the Brakes, a CTO Consulting podcast featuring Jan Esman and Jeroen Bolluijt.

Artificial intelligence is still widely framed as a risk problem. In boardrooms and governance forums, the dominant conversation centres on control: what to restrict, what to regulate, and how to prevent failure.

This framing, however, is fundamentally limiting. In the first episode of CTO Consulting’s Inside AI podcast, Esman and Bolluijt challenge this mindset. Using a race-car analogy, they argue that organisations are overly focused on the “brakes” of AI while neglecting the more critical question: how to see and navigate the track ahead.

From Control to Capability

Across government and enterprise environments, AI governance is typically implemented through layered control frameworks. Standards such as ISO, NIST, and local regulatory models are interpreted and operationalised at a granular level, often deep within organisational structures.

The result is predictable:

• Proliferation of controls

• Duplication across teams and systems

• Governance is embedded too far from business outcomes

While necessary, these controls are rarely sufficient to enable value.

As Bolluijt notes, organisations already have governance mechanisms in place. The challenge is not to rebuild control environments from scratch, but to determine what is genuinely AI-specific, and where existing frameworks can be reused or adapted.

This distinction is critical. Without it, governance becomes an exercise in accumulating controls rather than enabling value.

The Visibility Problem: Why AI Stalls

Despite significant investment in infrastructure, tools, and data platforms, many organisations struggle to move beyond pilots.

The constraint is not technology, but confidence.

Bolluijt frames this through a simple but effective analogy: a race car without visibility. Even with a powerful engine and functioning brakes, a driver who cannot see the track will hesitate or fail to start.

This lack of visibility manifests directly in enterprise AI environments:

• Unclear pathways from idea to deployment

• Fragmented use case management processes

• Limited understanding of how value is realised

In this context, governance alone cannot unlock adoption. What is required first is transparency.

A Three-Step Model for AI Adoption

The discussion introduces a practical operating model for AI enablement:

1. Transparency
Organisations must establish clear, visible pathways from use case ideation through to value realisation. This includes defined processes, decision points, and success metrics.

2. Confidence
Confidence is built through use. Teams develop trust in AI systems by engaging with them directly or observing successful outcomes from peers.

3. Value Realisation
Only once transparency and confidence are established can organisations consistently generate business value from AI investments.

This sequencing is critical. Skipping steps, particularly transparency, results in stalled adoption and underutilised capability.

Trust as a System, Not an Outcome

A key insight from the discussion is that trust in AI is not a static objective, but a dynamic system. Trust operates as a reinforcing feedback loop:

Use → Confidence → Trust → Capability → Value

This loop can reinforce positively or negatively. Poor early experiences reduce confidence, limit usage, and erode trust. Conversely, structured, low-risk adoption builds momentum.

For leadership teams, this reframes the challenge. Trust is not achieved through policy alone; it is engineered through experience.

AI as a Change Program

A critical observation is that most AI initiatives are not fundamentally technology programs.

They are change programs.

The introduction of AI alters workflows, decision-making processes, and service delivery models. The real value lies not in the model itself, but in how organisations adapt to use it.

This is particularly evident in sectors such as healthcare, where AI-enabled tools are introduced through iterative cycles:

• Experimentation

• Pilot deployment

• Operational integration

• Continuous improvement

The final stage, continuous improvement, is especially critical. Unlike traditional systems, AI capabilities evolve rapidly, requiring ongoing refinement to maximise value.

Invisible Governance: The Benchmark for Maturity

The discussion reframes what “good governance” looks like in the context of AI.

Rather than being visible and restrictive, effective governance should be:

• Embedded within workflows

• Aligned to business processes

• Largely invisible to end users

If governance introduces friction, it is misaligned.

This principle aligns with broader shifts in enterprise technology, where control mechanisms are increasingly abstracted and automated, allowing users to focus on outcomes rather than compliance processes.

Centralised Control, Decentralised Innovation

A critical architectural implication emerges from the discussion: the need to separate control from execution.

Leading organisations are beginning to explore:

• Enterprise-level AI controls (e.g. agent behaviour, model routing, policy enforcement)

• Decentralised experimentation at the business unit or end-user level

This federated model enables organisations to maintain consistency and compliance while accelerating innovation.

It also addresses a longstanding tension in enterprise IT: the balance between standardisation and agility.

By elevating controls to the enterprise layer, organisations can safely push experimentation closer to where value is created.

Strategic Implications for Leaders

For boards and executive teams increasingly asked how AI will transform their organisations, the implications are clear:

• Shift focus from control to capability

• Prioritise transparency over policy complexity

• Design for trust-building through usage

• Enable decentralised experimentation within enterprise guardrails

Most importantly, leaders must recognise that AI success is not determined by model performance alone, but by the organisation’s ability to adopt, adapt, and scale its use.

Making the Track Visible

The central thesis of the discussion is both simple and powerful: AI governance is not about applying the brakes; it is about making the track visible.

Organisations that focus solely on controlling risk will stall before they begin. Those who invest in transparency, confidence, and trust create the conditions for sustained value creation.

As AI capability rapidly commoditises, this distinction may ultimately define the leaders from the laggards.

About the Contributors

Jan Esman
Jan Esman is a seasoned digital advisory leader with extensive experience in IT strategy, enterprise architecture, and digital transformation. His expertise in aligning technology solutions with business objectives ensures that CTO Consulting clients benefit from strategic insights and effective digital initiatives.

Jeroen Bolluijt
Jeroen Bolluijt is an experienced AI and innovation leader with deep expertise in AI strategy, digital transformation, and emerging technologies. He focuses on translating advanced technologies into practical outcomes by aligning innovation initiatives with organisational priorities, governance, and delivery capabilities, enabling clients to realise measurable value from AI and digital initiatives.