The Elusive AI-Native Enterprise: How Much Ambition Do You Really Have?

AI-native is typically defined as a company (products, processes, …) that was designed with AI as the core component, not bolted on as a mere feature.

But as an existing company, how do you move towards an AI-native operating model?

For many organizations, AI is still treated as an adoption challenge:

• Which tools do we allow?
• Which use cases do we prioritize?
• How do we train our people?
• How do we govern the risks?

All valid questions. But not the full picture.

The real question is more fundamental:

Are we using AI to improve the way we work?
Or are we rethinking how the organization itself should work?

What is your ambition and did you think it through far enough?

That is why we developed the AI Ambition model. It challenges how far you should go. It helps you identify ceilings and pivot points, and how to prepare for them. And it gives you a clear metric, specific to your company and industry, to evolve towards.

The AI Ambition Model helps organizations define their AI target: the flag they choose to plant on the horizon. Defining how far, how broadly, how deep, and how autonomously they want to apply AI.

The AI Ambition model is build around four axes:

• Adoption: breadth and depth of use
• Autonomy: level of independence
• Agency: level of decision influence and outcome responsibility
• Process Coverage: how much of the work is handled by AI

How far you go on each axis, will depend on:

• External constraints: regulatory requirements, legal liability, sector-specific compliance
• Strategic choices: where you deliberately keep work human-led because it is part of your differentiation, trust, judgment, or relationship value
• Internal readiness: data quality, observability, security, adoption, cost, model reliability, organizational complexity (culture, change fatigue, …)

By drawing out your ambition, the model will also visualize where your company is at risk. It is not just a roadmap tool, it is a disruption lens. It shows where you are deliberately constrained, and where your organization is merely held back by legacy design.

Greenfield competitors will not have that legacy. They will design from day one with AI-native processes, AI-readable data, embedded feedback loops, and agent-ready systems.

This is not another maturity model.

A company does not become more mature simply by pushing every axis to the maximum. A legal department may deliberately keep autonomy low for high-risk decisions. An InfoSec team may allow broad adoption but tightly constrain agency. A marketing team may allow high autonomy in content variation but low agency in brand positioning.

Ambition defines the strategic direction.

Maturity is not “maximum AI.”

Ambition defines the strategic direction.

Maturity is not “maximum AI.”

Maturity is the capability to realize that ambition.

Adoption is often the first thing organizations measure. How many people use AI? How many teams have access? How many use cases are live?

But adoption has two different dimensions: breadth and depth.

Most companies start with measuring breadth, how many people are using AI. And while this is great as a first metric, and we would advice to start with this level, adoption depth will become more important as maturity and AI literacy grows.

Adoption depth measures how much of the work is being handled by AI. Some companies try to measure this through token consumption (tokenmaxxing), but they often find out that it was easily gamed resulting in cost instead of outcome. Instead we see this as work being performed within a value stream. Every use case not yet handled by AI is an opportunity for improvement.

A company can have broad but shallow adoption: many people use AI occasionally, but nothing fundamental changes. Another company can have narrow but deep adoption: one value stream is redesigned around AI and creates major leverage.

The AI Ambition Model makes that visible.

It prevents confusing tool rollout with transformation.

Autonomy describes how independently AI can execute work.

At low autonomy the AI is assisting and called on-demand. Most use AI in this way, through a conversation or by invoking AI to execute a specific task.

At high autonomy, AI will autonomously complete a goal.

Autonomy evolves across five levels, with two major shifts in human involvement. First, the human moves from making decisions (human-in-the-loop) to supervising (human-on-the-loop) the system. Later, at higher autonomy, the human moves out of the normal execution loop altogether (human-out-of-the-loop), while AI handles work independently.

Agency is often confused with autonomy, but they are not the same.

A robotic arm in a factory may have high autonomy: it can perform its movement without human intervention. But it has low agency: it does not decide what should be produced, what steps it will use to produce it, whether the production plan is still valid, or whether the business outcome is worth pursuing.

A high agency version would be robots moving between stations, improving how they build the product by changing the process, decide which product to build (red towels instead of black towels) based on market demands. Similar to how the physical internet enables just-in-time adaptability.

This is where organizations must be especially careful. Giving AI more agency is not just a technical upgrade. It changes accountability:

• Who is responsible when the AI acts?
• Who defines the guardrails?
• Who monitors drift?
• Who can override the system?
• Who owns the outcome?

The AI Ambition Model makes those questions explicit before the organization sleepwalks into them.

Many organizations begin with AI at the task level: write this email, summarize this meeting, generate this code snippet.

That is natural. People first use AI to improve the work directly in front of them, and tasks are the easiest place to start.

Over time, those individual tasks start to cluster into activities. A person does not only ask AI to write one email, but to prepare a meeting, analyze a document, create a first draft, or support part of a recurring activity.

But this is also where organic growth often reaches its ceiling.

The higher value of an AI-native organization does not come from isolated task improvement alone. It comes from orchestrating consecutive work across workflows, processes and eventually full value streams.

This requires a deeper transformation by investing in upskilling, role redesign, accountability, controls and operating model clarity.

Done well, governance helps break through this ceiling.

In software development, the rise of roles such as the product engineer and the forward-deployed engineer illustrates this shift. As AI takes over more of the task execution, human value moves toward problem framing, orchestration, validation, context ownership and outcome responsibility.

Organic AI usage is bottom-up, and grows across all 4 axis.

It spreads through curiosity, convenience and individual productivity. That is where most companies begin, and there is nothing wrong with that.

Organic growth creates champions, a coalition of the willing.

But on every axis there is at least one tipping point, where organic growth stagnates and transformation and governance are required to evolve towards your AI-Native Ambition.

Transformation requires deliberate redesign.

At some point, the organization must ask:

Which workflows should be redesigned?

• Which decisions can be delegated?
• Which controls must become embedded?
• Which roles need to change?
• Which systems must become queryable?
• Which operating model assumptions no longer hold?

Without that, the company has adopted AI tools, but the work still flows through old structures. People still copy information between systems. Decisions still depend on meetings. Processes still break across silos. Dashboards still describe the past instead of becoming actionable interfaces. Governance still happens after the fact.

This is why AI-native cannot have one universal definition.

A bank, a hospital, a software company, a government agency and a logistics firm should not plant the same flag.

Their risk profiles differ.
Their regulation differs.
Their data maturity differs.
Their operating models differ.
Their trust boundaries differ.
Their appetite for autonomy differs.

For one company, AI-native may mean broad adoption with low agency: everyone is augmented, but decisions remain human. For another, it may mean deep autonomy in a few operational processes: narrow scope, high automation. For another, it may mean becoming a queryable organization first: making knowledge, processes and systems legible before delegating meaningful work to AI.

For a greenfield disruptor, AI-native may mean designing the company around AI from day one.
For an established company, it may mean selectively redesigning value streams while protecting the stability of critical operations.

The AI Ambition Model gives companies a language to make those choices explicit.