Stop trying to work faster; speed up the work

Everyone gets Copilot. Customer service gets a chatbot. Marketing creates content with ChatGPT. And from now on, the finance department uses AI to summarise invoices. Look at us innovating! Or can we do better than that?

Don’t get me wrong: if AI helps an employee complete a task in ten minutes instead of thirty, that is a clear win.

But it is not the huge productivity leap we have been promised for the past few years.

Because while everyone may be working a little faster, the work itself often moves through the organisation at exactly the same pace. The quote is still waiting for approval, and the order still has to pass through four different departments. We use AI to speed up individual tasks while leaving the surrounding process untouched.

A missed opportunity, if you ask me.

A real-world example

Take an everyday business process such as order intake. An order arrives as an email attachment. Someone opens the file, looks up the customer, checks the product numbers, enters the details into the ERP system and sends a confirmation.

AI can make almost every part of that process a little faster. AI reads the email. AI summarises the order. AI helps find the right product number. AI drafts the confirmation.

Fantastic. Except there is still an employee sitting between every step, checking, copying, pasting, clicking and nudging the process forwards again.

You have sped up the individual actions, but not the work itself.

So the real question is not: How can AI help this employee work faster?

The question is: Why does an employee still need to carry out this process step by step in the first place?

Personal productivity is not the same as business productivity

Many AI strategies begin with personal productivity. That makes sense: the benefits become visible quickly, and implementation is relatively straightforward.

The problem is that individual time savings do not automatically add up to a faster company. If ten employees each save half an hour, but the process still takes three days because of handovers, waiting times and checks, what you mainly have is ten slightly less busy employees.

A perfectly decent result, but still small beer 🍺.

Real business productivity emerges when the entire process contains fewer actions, handovers and delays. That is the difference between working a little faster and speeding up the work itself. The first delivers percentage gains. The second completely transforms your end-to-end lead time.

Your processes were designed for another era

Many of today’s business processes were designed in a world without AI agents. A world in which people had to find information, assess it and transfer it from one system to another.

Our processes are built entirely around that reality, including handovers and roles that exist primarily because system A cannot communicate with system B on its own.

But that world has changed. An AI agent can gather information, understand context, make decisions and independently perform actions in your business systems.

You do not give an agent a prescribed sequence of clicks. You give it a goal, context, tools and clear guardrails. The agent then determines which steps are needed. That is fundamentally different from using AI to push an employee through an outdated process a little faster.

An office is a decision factory

The average office-based organisation may not produce boxes or machines, but it does produce decisions.

For every decision, people gather information, assess the context and take action, often across multiple applications and departments. That production line has evolved around what people can handle: the human scale.

AI agents are not constrained by that scale. They can assess information continuously and interact instantly with other agents and systems. A person only needs to enter the picture when an exception, risk or meaningful choice requires human attention.

But you do need to design your digital production line accordingly.

Let agents do the work

Imagine a customer calling about a delayed order. In a traditional organisation, several employees may check the order status, stock levels, delivery options and possible compensation.

With an AI agent, this can become one continuous process. The agent retrieves the order, checks the stock, calculates a new delivery date and updates the CRM. An employee is only brought in when something falls outside the agreed parameters.

Not to carry out the entire process manually after all, but to make the one specific decision that genuinely requires human judgement.

That is the point at which AI no longer merely assists with the work, but actually performs it. And that is when the pace of change really takes off.

Apply that principle to more processes and a fundamentally different kind of organisation emerges: the Autonomous Enterprise.

An organisation in which people and AI agents form a single digital team. Agents execute processes largely independently, across systems and departmental boundaries. People step in when their knowledge or judgement is truly needed.

Design for automation

If we want to unlock the real value of AI, we need to stop sprinkling AI over isolated tasks.

You do not start with the existing workflow and ask where AI might fit into it. You start with the desired outcome and work backwards to design the smartest and safest route.

What information is needed? Which decisions can an agent make independently? Which systems must it be able to interact with? When is human approval genuinely necessary? And which steps exist only because that is how we have always done things?

That last question is where it gets interesting. A surprising amount of work turns out to be little more than the glue holding together outdated systems, departments and procedures.

If agents can connect those systems directly, you can simply remove the glue.

Questions for the boardroom

The journey towards an Autonomous Enterprise requires more than a collection of smart agents. We are at the beginning of a much bigger transformation than the introduction of yet another generation of software.

This is the Great Re-engineering of business operations.

Processes that once made perfect sense need to be reconsidered. Not because they were poorly designed, but because the assumptions on which they were built no longer hold true.

That is precisely why the Autonomous Enterprise is not an IT project.

Technology, data, integrations and architecture are obviously crucial. An agent without reliable data or access to your core systems is little more than an enthusiastic intern without a laptop.

But the most important questions are organisational. Who owns a process that cuts across multiple departments? Which decisions do we want to automate? Where does human oversight remain essential? And are we prepared to redesign roles, responsibilities and KPIs?

These questions therefore do not belong solely to IT or innovation. They belong in the boardroom.

People are not disappearing. The robotic work is.

The word ‘autonomy’ quickly conjures up the uncomfortable image of an organisation in which people have become redundant. That is not what I envisage.

I see an organisation in which people no longer spend their days moving, checking and re-entering information. One in which employees focus on exceptions, relationships, creativity and judgement.

This is not a story about people or AI.

It is about one digital workforce in which people and AI agents work together and suddenly complete a month’s worth of work in a single day.

Not because people work harder, but because the work is organised in a fundamentally different way.

Welcome to the Autonomous Enterprise

The Autonomous Enterprise is not an organisation that automates everything once and then considers the job done. It is a new way of operating.

An organisation in which processes run largely autonomously, agents collaborate across departmental boundaries and people retain control where it matters.

It does not start with a hundred agents and a sweeping reorganisation. It starts with one carefully chosen process. A process in which manual work, waiting times and handovers are currently slowing everything down, and where there is enough business value to justify a serious redesign.

So don’t ask which AI tool you should buy. Ask which work you are prepared to reinvent.

Because if you use AI to make isolated tasks a little faster, you end up with isolated tasks that run a little faster. If you redesign entire processes around people and agents, you build an organisation capable of achieving more than was previously possible.

To me, that is the real promise of AI.