Designing AI-Native Organisations

A manufacturer learns that a shipment of components will arrive late. Its production planner uses AI to prepare a revised schedule. Procurement identifies an alternative supplier, while sales considers which delivery commitments can change. Each department can reach an answer more quickly than before. The difficulty is agreeing which answer the business should act on.

Buying replacement components may protect delivery dates but reduce the margin on an order. Changing the production schedule may preserve that margin while delaying another customer. Better information helps, but someone still needs to resolve the competing priorities.

In our previous article (opens in a new tab), we examined how to define work before assigning it. Once those responsibilities are clear, a broader question follows: how should the organisation be arranged to carry them through to an acceptable outcome?

Give the outcome an owner

Departments provide useful homes for specialist knowledge. Procurement understands suppliers and commercial terms. Production understands capacity and manufacturing constraints. Finance understands the consequences of different spending decisions. Those capabilities remain valuable when AI participates in the work.

The problem arises when responsibility ends at the departmental boundary. Procurement may have completed its assignment by finding another supplier, even though nobody has authorised the additional cost. Production may have produced a workable schedule that sales cannot reconcile with existing commitments. Every contribution can be reasonable while the overall result remains unresolved.

Organising across these boundaries is not a new idea. The UK Government Service Standard, for example, calls for multidisciplinary teams and includes decision-makers so that teams can respond to what they learn. AI does not invent the need for this arrangement. GOV.UK Service Standard (opens in a new tab)

What AI changes is how much of the underlying execution may be reassigned. Preparing a production schedule could move from manual planning to conventional optimisation software, with AI helping interpret changing circumstances. Responsibility for meeting delivery commitments still needs a clear home.

For the manufacturer, that might mean an accountable owner for order fulfilment, working with the relevant specialists. This person needs agreed decision rights and access to resources. Giving someone responsibility for delivery without any influence over the decisions affecting it creates an escalation point rather than effective ownership.

Make the organisation support the decision

An organisation chart will not resolve the problem if budgets and performance measures pull people in different directions. Suppose procurement is rewarded primarily for reducing purchase prices, while production is measured on keeping machines busy. Both teams may resist an expensive, disruptive change that is nevertheless the best way to fulfil an important order.

AI can make those local priorities more effective. It does not determine whether they are sensible for the business as a whole. An agent optimising purchasing costs can reproduce the same conflict as a department doing so, only with less delay between decisions.

Designing an AI-native organisation therefore includes examining what people are authorised and encouraged to do. The fulfilment owner might have permission to approve additional spending within agreed limits. Larger changes would need a defined decision route, with someone available to make the decision. Performance should reflect the completed business outcome, including its cost and quality.

There are limits to combining responsibility. The person accountable for delivery should not be able to waive a safety requirement simply because an order is urgent. Independent quality assurance and separation of duties can be deliberate protections. Bringing work together does not mean removing every boundary.

Nor does every outcome require a permanent team containing every specialist. Shared expertise may be more practical, particularly where demand varies. What matters is whether the arrangement gives people a workable way to act together, including when their priorities conflict.

Keep expertise within reach of the work

As execution changes, the responsibilities within a team may change too. A planner who previously assembled schedules might spend more time testing assumptions and investigating why apparently feasible plans fail on the factory floor. Procurement expertise may become more important when evaluating alternatives that software can identify but cannot adequately assess.

That transition deserves more attention than a new job title. People need time to learn, access to the underlying evidence and a meaningful say in how their work develops. Some responsibilities may disappear. Others may become more demanding because routine cases no longer occupy much of the working day.

There is also a continuity problem. If experienced staff only handle exceptions, while newer colleagues never encounter ordinary cases, the organisation needs another way to develop practical judgement. Supervised work and opportunities to examine real decisions belong in the operating model.

Keeping people accountable while removing their opportunity to understand the work is a fragile arrangement. Human oversight has value when people can exercise informed judgement and intervene effectively.

Start with a change the business can judge

For an established organisation, this need not begin with a company-wide restructuring. The manufacturer could start with its response to delayed components. It could examine how a disruption is identified, how alternatives are assessed and who can authorise a revised plan. Existing applications and teams may already support much of that work.

This is where NativeWork provides an architectural starting point. Describing the work separately from its current executor makes it possible to examine a change without assuming that every existing departmental or application boundary must remain intact. The organisational design then needs to support the responsibilities that emerge.

The first improvement might be AI-assisted assessment of alternative supply arrangements. It might also be a simpler approval route or the removal of a duplicated check. The test is whether the business reaches a sound decision sooner, with acceptable costs and consequences. Faster preparation of proposals is only part of that result.

Evidence from that bounded change can inform the next one. Some responsibilities may stay where they are; others may need different ownership or resources. Continuous Enterprise Evolution requires room for those organisational adjustments alongside technical improvements.

The late shipment will still present the manufacturer with difficult choices. An AI-native organisation should be better equipped to make and carry out those choices because its knowledge, authority and execution work together. Faster answers are useful when the organisation can turn them into a coherent decision.

About NativeWork

NativeWork™ is an AI-native Enterprise Architecture standard and method developed by Centipod. It places work at the centre of enterprise design and provides an architectural foundation for Continuous Enterprise Evolution.

WorkML™ is the accompanying language for expressing NativeWork architectures in a form designed for both people and intelligent systems.

Learn more at nativework.org (opens in a new tab).

NativeWork™ and WorkML™ are trademarks of Centipod. Trademark applications are pending.

This article was created by people. We have used artificial intelligence (AI) to help articulate our message and refine the text. AI was employed as a tool to assist with structuring, identifying grammatical and spelling errors, and improving readability. The final document has been carefully reviewed and approved by our team.