Author(s): Bram Nauts

Originally published on Towards AI.

AI will not transform because it’s deployed – it will transform because the way of operating is redesigned. The tricky part? Transformations rarely fail at the start, they fail in the middle – when organisations try to scale.

In a previous article I defined the concept of the AI-native bank. A bank where decisions, processes and customer interactions are continuously driven by AI. Since publishing that article, one question came up repeatedly: “How do we actually get there?”

Before exploring that question, it is important to acknowledge something. The idea of AI-native organisations is still largely a promise. The potential of AI is enormous, but the long-term economics and risk profile of AI-driven companies are still emerging. Some initiatives will deliver extraordinary value. Others will fail to scale.

But despite this uncertainty, one thing is becoming increasingly clear. The opportunity is too large to ignore. Across industries, AI is beginning to reshape how companies operate. Technology firms are embedding AI into decision-making. Digital platforms are automating complex processes. New entrants are building organizations designed around AI from day one.

If this shift continues – and all signs suggest it will – the competitive landscape will change dramatically. Companies that operationalize AI effectively will unlock: faster decision cycles, lower operational costs, more personalized customer experiences and changing business models.

Companies that move too late risk something more dangerous than short-term inefficiency. They risk becoming structurally slower than their competitors. The real question therefore is not whether they should experiment with AI. They already are. The real question is: “How do organisations move deliberately toward becoming AI-native?”. Therefore in this article I will outline the best practices I apply:

1. Create a plateau plan to guide and lay out the journey, and;

2. Apply transformational best practices to navigate change.

The Plateau Journey Toward an AI-Native Organisation

Transformation in general don’t progress smoothly. Organisations move through plateaus. At each plateau the organization evolves: adoption, foundation and value creation. Understanding these plateaus helps leaders identify where they are – and what must happen next.

Plateau 1 – Exploration and Foundation

Most organizations begin their AI journey through experimentation. Teams explore use cases such as document processing and internal productivity copilots. The goal is learning.

Scope of this plateau at a minimum

• A set target of use cases

• A top-down cost reduction target to start embedding value

• AI Governance aimed at AI Act compliance, key internal and external risks and increasingly literacy

• Testing scalability of the Data & AI platforms and other foundations

The kill list

• Stop any use case without business owner and contribution to the value case – ensure strategic alignment

• Stop any shadow AI – get in control

What success looks like

• pilots deliver measurable improvements

• teams gain confidence in AI

• employees begin using AI

• leadership recognises strategic potential

• Foundation capabilities are tested and their improvements areas are clear

Typical bottlenecks

Experimentation quickly exposes structural issues:

• fragmented data leading to AI using ungoverned data

• unclear roles & responsibilities delaying decisions and misaligning priorities

• limited AI literacy hampering true adoption of the AI use cases

KPIs to set and track

• high-impact use cases in production

• Time to production

• % of total population related to the use case using the AI use case

• #AI risks identified, incidents & ethical

• #Lessons learned implemented

Leadership question to be answered: Where are we seeing real value from AI – and which experiments should become strategic priorities?

Plateau 2 – Strategic Verticalisation

The second plateau begins when organisations stop asking: “Where else can we experiment with AI?” and start asking: “Where can AI fundamentally transform our business?”. Investment concentrates on a few high-impact areas. In banking these often include:

• customer servicing

• financial crime and KYC

• credit and investments

• operations

Scope of this plateau

• 3–5 Holistic value areas to deploy AI end-to-end across disciplines

• Modernised data & AI platform focussed AI-ready-data (e.g. investing in knowledge graphs and vector databases)

• Explicit governance with guardrails focussing on accelerating and controlling what matters

The kill list

• Stop batch decisioning (overnight risk & fraud) and manual case handling – move to realtime to harness the benefit of AI and dare to stop the ‘old way of working’

• Freeze all unrelated use cases – move top down on the values areas – and focus your talents and experts where it matters

• Stop an experimentation setting and demand AI systems to be monitored and promote sharing learnings

What success looks like

• entire journeys related to the value areas redesigned around AI

• faster decision cycles

• improved customer experience

• meaningful cost reduction

Typical bottlenecks

• Central AI and data platform constraints hampering fast deployments

• Unclear AI governance, clear on paper but not harnessed in practice

• Lack of alignment between business and technology priorities hampering speed

Leadership question to be answered: Which few domains should we transform with AI – and are we focusing our investments strongly enough there?

Plateau 3 – Enterprise-Scale AI Acceleration

Once AI becomes critical across multiple areas, the organisation must evolve its ability to deliver AI at scale. AI becomes a repeatable enterprise capability. Equally important is solidifying the foundation, and with that put focus on transforming leadership and the workforce. Employees must learn how to work with AI systems and leaders need to become extremely bold to push change.

What success looks like

• AI solutions move rapidly from development to production

• AI systems are continuously monitored and improved

• AI-ready-data products are reused across teams

• AI becomes part of daily operations

The kill list

• Process-and-system based operating model – organise around AI

• Product-and-discipline centric silo’s – organise around AI

Typical bottlenecks

• resistance from the workforce

• unclear and not formalised risk appetite

• uncontrolled “citizen AI” experimentation leading to more risks

Leadership question to be answered: Can our organization reliably and responsibly deliver and scale AI solutions?

Plateau 4 – AI-Native Operations

At the final plateau AI becomes embedded in the way of operating. Customer journeys are orchestrated through intelligent workflows. Decisions are supported or automated in real time. Employees increasingly work alongside AI systems rather than performing routine decisions themselves.

What success looks like

• AI embedded across core operations

• faster decision cycles

• adaptive data & AI driven processes

• structural value creation

The company becomes AI decision-driven rather than system-driven.

Leadership question to be answered: How should our operating model evolve if AI becomes the core decision engine of the bank?

What Previous Transformations Teach Us

The idea of AI-native organisation may feel unprecedented. But many companies have navigated major transformations before. And from there is one core lesson learned: technology is not the real change. Same with the shift toward AI-native it is not simply a technology rollout. It is an organisational transformation – and history shows that such transformations only succeed when leadership treats them as fundamental change.

Research on organizational change, most famously the work of John P. Kotter, consistently shows that successful transformations follow several principles: creating urgency, aligning leadership, removing structural barriers, generating early wins, and embedding new ways of working into the organization. These lessons are visible in major corporate transformations.

IBM reinvented itself in the 1990s under Lou Gerstner, the company did not simply adopt new technologies. It reorganized around services, broke internal silos, and forced leadership alignment around a new operating model.

More recently, Microsoft’s cloud transformation under Satya Nadella required redefining strategy, changing culture, and aligning the entire organization around a new platform model.

And Netflix’s evolution toward a data-driven company required embedding analytics and algorithms into core decision-making across the business.

The lesson across these transformations is clear: Technology does not transform organizations. Leadership decisions do. AI transformation will require the same discipline – but the challenge may be even greater.

The transformation playbook

Unlike previous transformations, AI has the potential to reshape the decision-making horizontal and vertically through the organization: risk assessment, customer interaction, operational workflows, and strategic insights. That means AI cannot remain a collection of local initiatives inside individual departments. It must become a top strategic priority for the entire institution, because it cuts across department and discipline. Therefore, I can consolidated a short but powerful transformation playbook based on best practices.

1. Create real strategic urgency by being bold and pauzing other initiatives

Executive leadership must explicitly prioritise the AI transformation. If AI competes with dozens of other initiatives, it will remain incremental and never breakthrough as it will shake the core processes through the bone. If other change initiative aren’t deprioritised you can’t change your full credit to risk management process. Therefore, real transformation requires elevating AI to the top of the strategic agenda – and sometimes pausing or stopping other initiatives that dilute focus. Often this is where it gets stuck, AI is added to the change agenda but given its transformative power it comes to a grinding halt when leaders try to scale. It remains spot-based instead transforming the full value chain and operating of it.

Lessons from other transformation shows that siloed ownership must be broken. Many organisations attempt to scale AI within individual departments – which can be fine depending where you are in your journey. However, if you are in the later plateaus it needs cross-domain change: data & AI platforms, customer journeys, risk & governance processes, and operations must evolve and transform together.

2. Remove foundational barriers to gain execution speed

We shouldn’t forget that starting an AI journey will lead to many bottlenecks to overcome. The AI platform will be a bottleneck. Governance will slow down. Central AI activators are not pushing hard enough. Risk and HR processes didn’t adapt to AI. Without clearly recognising that foundational aspects matters, initiatives stall in organisational friction. This means that the foundation that AI transformations lean on needs to evolve fast in line with the adoption pace. Often focus and applaus goes to the development of AI – but leaders shouldn’t forget the building blocks to get there.

3. Generate wins and momentum to show what is possible

Organisations must generate visible wins early. Early successes create credibility and build momentum – a core principle in Kotter’s change framework. People working in the organisation – and especially the typical laggers on AI – need wins to start believing. Without early proofpoints – like plateau 1 pilots – transformations will slowly die. Therefore, in transformation roadmaps don’t underestimate the power of a communication plan (internal and external). Important lesson is that success doesn’t stop with deploying AI it needs the adoption before celebrations start. So the impact is seen and felt, otherwise it remains a gimmick.

4. Embed periodic reflection moments to steer

Finally, major transformation share one characteristic. The roadmap toward success is rarely straight. Unexpected bottlenecks appear. New opportunities emerge. Assumptions change. AI transformation is even more unpredictable because the technology itself is evolving rapidly. That means the journey cannot rely on a rigid roadmap. Instead, organisations must establish structured reflection cycles anchored in the plateau plan. At regular intervals – typically every quarter given the pace of this fast moving technology – leadership should step back and ask:

• Where are we on the plateau plan?

• What is the real return on investment of AI?

• Which bottlenecks are slowing progress?

• What fundamental capabilities must we strengthen next?

• Are we seeing change in the core of our company?

• Do we have the talent and workforce to take the next step?

Without this discipline many organizations fall into a familiar trap. AI initiatives multiply. But structural progress remains limited. With reflection and recalibration, they can continuously adjust their course. Because the journey toward AI-native is not about executing a perfect plan. It is about learning faster than the challenges and technology emerges.

Published via Towards AI