Business redesign & workflows

Do leaders need to move from asking where AI fits into the existing business to asking how AI changes the business itself?

Do leaders need to move from asking where AI fits into the existing business to asking how AI changes the business itself?

Video · 5 min read

The short answer

Ask both where AI can improve existing work and whether it changes how that work should be organised. Follow one process across teams, examine what each handover contributes and define when the result is genuinely complete. Redesign around the outcome while preserving necessary judgement, challenge and approval.

Ask both where AI can improve existing work and whether it changes how that work should be organised. Follow one process across teams, examine what each handover contributes and define when the result is genuinely complete. Redesign around the outcome while preserving necessary judgement, challenge and approval.

Watch: Do leaders need to move from asking where AI fits into the existing business to asking how AI changes the business itself?

Watch the explanation, then explore the practical implications below.

Improve the task and question the wider process

A local productivity improvement can be useful without resolving the overall delay. One team may prepare information faster while the next still waits for clarification, approval or access to the right records. The business outcome depends on how these parts connect.

Begin with an end-to-end result, such as an agreed quotation, a usable project handover or a resolved customer enquiry. Then ask which activities are necessary to produce it and which exist because information has historically been difficult to move or interpret.

Follow one piece of work across the boundaries

Choose a recent example and trace it from the initial request to the final result. For every stage, identify the owner, inputs, decisions, waiting time and output. Ask the receiving person whether they had what they needed to proceed.

Pay attention to the moment one team considers its work finished. Sending a document may complete their task without making the next person’s job possible. Missing context, unclear status or inaccessible evidence can create another cycle of questions.

Separate useful challenge from unnecessary handovers

Some stages exist to provide independent review, specialist judgement or authority to make a commitment. Removing them without understanding their purpose can weaken the process. Other stages may consist mainly of retyping, formatting or collecting information already held elsewhere.

Label the purpose of each handover before redesigning it. Decide what evidence and approval the outcome requires, then consider whether the same control can be achieved with less duplication. Faster movement is useful only if the recipient can trust and use what arrives.

An example: a tender-to-delivery handover

A bid team may assemble requirements, a commercial team may agree assumptions and a project team may later inherit the commitment. If exclusions, changes or unresolved questions are lost between them, faster drafting does not solve the problem.

A redesigned workflow could maintain a shared, permission-controlled record of requirements, approved decisions and unresolved issues. AI may help organise evidence and prepare a handover summary. The receiving team still confirms that the information is sufficient, and authorised people approve changes to commitments.

Define completion around the recipient’s needs

Agree what must be true before the workflow is considered complete. For a handover, that might include an approved scope, source references, named owners for open issues and acceptance by the receiving team. Use criteria appropriate to the actual work.

Measure the whole journey: elapsed time, clarification loops, rework and the quality of the result. If one department becomes faster while another absorbs extra review, the improvement may simply have moved the effort. Make that trade-off visible.

Give the cross-team process an owner

A workflow that spans departments needs someone who can coordinate changes across their boundaries. Local teams should contribute their expertise, but the overall outcome cannot depend entirely on each department optimising its own task.

At CSM, we would start with one recurring process, map the information and decisions with the people involved and agree a bounded redesign. We would test it with real cases, retain necessary approvals and assign maintenance responsibilities. The aim is a connected way of working that improves the finished result, rather than faster departments joined by the same old delays.

Questions for your leadership team

What finished result are we trying to produce?

Define the outcome from the customer’s or receiving team’s perspective.

Where do handovers create clarification or waiting?

Follow a recent example and ask each recipient what was missing.

What purpose does each stage serve?

Distinguish judgement, challenge and approval from copying, formatting or unnecessary coordination.

Who owns the process across departments?

Name someone able to coordinate changes and resolve competing local priorities.

Are we measuring the whole journey?

Track completion time, review, rework and usefulness to the recipient, not only departmental speed.

Full video transcript

The full video transcript is included below for reference.

I think you have to ask both, but you can’t stop at the first one. Once AI can carry out a meaningful part of a task, you start running into questions about how the organisation is put together.

Jobs start to blur. Someone who understands a customer problem can do more of the work of exploring a solution. Someone in operations can investigate information they previously needed another department to prepare. People can move further before they have to hand the work over.

And that matters because a lot of our processes are built around those handovers. One person gathers the information, another analyses it, another turns it into a presentation, and then someone else makes the decision. Some of that exists for a good reason. Some of it exists because that was the only practical way to get the work done.

So you need to ask what each stage is contributing. Is this a necessary judgement? Is it a useful challenge? Or are we passing information between people because our systems don’t let them work from the same information?

The danger is that you put AI into every department and each department gets faster at its own part, but the work still gets stuck in exactly the same places. You’ve improved the individual jobs without looking at the relationship between them.

That brings you back to quite fundamental questions. What are we trying to produce? Who is it for? What needs to happen before we can say the work is actually finished?

You don’t have to redesign the entire company to start answering that. Pick one process that crosses a few teams and follow it all the way through. The interesting questions usually appear at the point where one person thinks they’ve finished and the next person still doesn’t have what they need.

Put these questions to work in your business

Bring a real decision or workflow to a conversation with CSM. We can explore the evidence, responsibilities and practical steps needed to make AI useful for your team.