Don’t Rush to Replace Your Team with AI

Before removing a role, prove the new process works—and consider what an experienced person could do with better tools.

At Net Age, we use AI every day. We spend tens of thousands of rand on AI tokens, build technology with AI assistance and continue exploring what these tools can do for our business and our clients. We also still have designers, developers and people who speak directly to our customers.

Those decisions belong together.

Our experience of AI has made us more interested in what people can achieve with it, and more cautious about assuming that a convincing demonstration means an entire role can disappear. A tool completing a task is useful evidence. It is not, on its own, evidence that the business no longer needs the person who used to perform that task.

That is our message to business owners considering changes to their teams: slow down long enough to understand what you are replacing. Test the complete process, include the difficult work and give experienced people a chance to help shape what comes next.

You may discover that the biggest opportunity is not removing them. It may be changing how they spend their time.

We Understand Why the Idea Is Attractive

When you first see AI produce a design, write content or generate working software, it is natural to compare that speed with the time a person would have taken.

For a business owner responsible for costs, delivery and growth, the possibilities can look compelling. Something that once required a specialist now appears to happen after a few instructions. It is easy to jump from that observation to a much larger conclusion: perhaps the specialist is no longer necessary.

We understand that reaction because we have experienced the excitement ourselves.

At Net Age, using AI to build technology has reinforced the importance of experienced people. Producing a first version is only one part of delivering work that a business can rely on.

But the more we built, the more we encountered work beyond the first visible result. Our development team still had to help examine and address issues with security, database structure, SEO and the reliability of what had been produced.

Being able to create something quickly did not remove the need to know whether it was right.

A Task Is Only Part of a Role

Before deciding that AI has replaced a position, write down what the person actually contributes.

A designer may create artwork, but the work can also involve understanding a brief, interpreting feedback, recognising an inconsistency and knowing when the requested change will weaken the result. A developer may write code while also investigating failures, considering dependencies and maintaining systems that other people rely on.

Those examples will differ between businesses. The point is to examine the real role rather than its most visible output.

Ask what happens before the task begins. Who clarifies the request? Who notices that the supplied information is incomplete? Then look at what happens afterwards. Who checks the result, explains a decision, responds to a problem and takes responsibility for correcting it?

If those activities remain necessary, they have not disappeared simply because AI produced the first version. They may have moved to another person, including the business owner.

A staffing decision that overlooks that transfer can make a process appear cheaper while quietly creating work somewhere else.

We Still Need People Who Can Judge the Result

At Net Age, using AI does not mean handing over the final judgement.

We still need people who understand the client, the purpose of the work and the standard it must meet. We still need someone who can recognise when an answer sounds plausible but does not solve the actual problem.

That distinction is particularly important when a business owner is assessing work outside their own area of expertise.

An attractive interface does not tell you everything about the software underneath it. A polished design does not automatically tell you whether it follows the brief. Fluent writing does not establish that its claims are accurate or that it communicates what the client intended.

AI can contribute to all of these activities. The business still needs a dependable way to evaluate the contribution.

Removing the person who can make that judgement before establishing an alternative is a serious decision. It should be supported by evidence from real work, not excitement about what happened in a demonstration.

Test the Process Before Making an Irreversible Decision

Our advice is to run the new way of working alongside the existing one for a meaningful period.

For a substantial change to an established role, consider a year of real operating experience before using the results to justify removing the position. That is a deliberate recommendation to test across different conditions, not a claim that twelve months is a magic number or a requirement for every small automation.

The period should be long enough to include busy weeks, unusual requests, revisions, staff absences and the jobs that do not follow the usual pattern.

Start with a defined set of tasks. Record how the current process performs, then compare the AI-assisted process against it. Include time spent preparing instructions, reviewing output, correcting mistakes and handling exceptions.

A faster first draft is worth knowing about. So is a slower approval process. If the person checking the work repeatedly has to rebuild it, that belongs in the comparison.

The question is whether the complete process delivers acceptable work consistently, with a clear owner when something goes wrong.

Measure What Matters to the Business

Speed and subscription fees are easy to see. They are only part of the assessment.

Look at the time from receiving a brief to delivering approved work. Record the amount of rework, whether deadlines are met and how often someone has to step in. Consider whether clients are satisfied with the result and whether the team can explain or reproduce it.

Also look at capacity. If a designer can now complete routine work faster, what happens to the time saved? Is it used for more valuable design work, better quality control or additional client projects? Does the benefit appear in the business, or is it absorbed by an increasingly complicated process?

Include the cost of training, integration, oversight and ongoing maintenance when those apply.

The objective is not to make AI look expensive or to defend every existing task. It is to compare two real ways of operating. You cannot make that comparison fairly if the human process includes all its responsibilities while the AI process is measured only on its best output.

Today’s Price Is Not a Permanent Business Plan

We would also be careful about making long-term staffing decisions around a tool’s current price.

That does not require a prediction that every AI service will become more expensive. Some costs may fall, others may rise, and the features a business needs may change. The practical issue is that a price available today is not a promise about every future version of the service.

Build a few scenarios into the decision.

What happens if usage grows faster than expected? What if a cheaper model needs more supervision than the one used in the trial? Could the business continue operating if access changes or the chosen service is temporarily unavailable?

A useful process should have an answer beyond hoping that the original assumptions remain true.

We spend heavily on AI because it can support valuable work. But a budget decision should consider the total cost of delivering that work, rather than treating tokens or a subscription as the entire cost of replacing a person's contribution.

Your Experienced Team Has Knowledge Worth Using

People who have worked in a business for years carry knowledge that may not be captured in a formal brief.

They know which client needs a careful explanation, which request is likely to expand and which seemingly small detail has caused trouble before. They may recognise a problem early because they have already dealt with its consequences.

That knowledge deserves a place in the transition.

If technology reduces the time needed for one part of a role, talk to the person about where their experience could be useful next. Could they take more responsibility for reviewing work? Could they improve the process, support customers or help others use the new tools effectively?

The answer will not be the same for every person. Moving someone from task A to task B requires a genuine role, appropriate training and realistic expectations. It should not amount to telling them to work out their own future without support.

At Net Age, that is the direction we want to pursue: work with the people who have helped build the business and find how new capabilities can make better use of their time.

Involve the Team in Testing the Tools

The people doing a job are often well placed to identify what a trial needs to include.

Invite them to bring the awkward briefs, the recurring exceptions and the examples that usually require several rounds of correction. Those cases are valuable because they reveal what the proposed process can handle and where it still needs help.

Give the team permission to report failures honestly. If every concern is treated as resistance to change, you risk losing the information needed to make the change work.

Equally, the trial should be open to genuine improvements. Familiarity with an existing process is not a reason to keep every step unchanged. Some work may no longer need to be done in the same way.

The aim is to learn together which tasks can be automated, which can be accelerated and which still require experienced attention.

That produces a more useful outcome than asking staff to prove either that AI is useless or that their role can be eliminated.

Clients Still Need Someone to Speak To

Our clients do not simply receive access to an AI and get left to work everything out.

We talk to them. We discuss what they are trying to achieve, look at the work and help make decisions about what should happen next. AI supports parts of that process, but it does not remove our responsibility for the relationship.

That is a business choice, and one we consider important.

A client may need more than a quick answer. They may need someone to understand a concern, explain a trade-off or take ownership when the result falls short. The person handling that conversation needs context and the ability to act.

When considering a change to your own team, ask what your customers value about dealing with your business. Which of those expectations will remain, and who will meet them after the change?

Efficiency is useful when it improves the service. It deserves a closer look when it removes the support that made the service worth buying.

Change Roles Deliberately

None of this means that jobs and business structures will remain unchanged.

There may be tasks you no longer need, services you can deliver differently and new responsibilities that did not exist before. A business should be able to respond to those possibilities.

Our argument is about the sequence of decisions.

First establish what the technology can reliably do in your circumstances. Then work out what human responsibility remains and how the team can develop around it. Make staffing decisions with that evidence in view.

A simple starting point is to choose one process, appoint a responsible person, agree the standard and run a measured trial. Review the results regularly. Keep a route back if the new process fails to meet the business's needs.

This allows you to move forward without pretending that a successful experiment has answered every operational question.

AI Should Help Your People Do Better Work

We are enthusiastic about AI because we use it and see the opportunities it creates. We are cautious about replacing people too quickly for exactly the same reason.

Our experience has shown us how much can be built, how quickly confidence can grow and how much work can still remain after a tool says something is finished.

Before you remove an experienced person from your business, make sure you understand the full contribution you are removing. Test the alternative in ordinary work and difficult conditions. Account for oversight, changing costs and the relationships your business depends on.

Then ask what that person could achieve with the right tools and support.

You may find a better way to organise the work. You may create capacity for growth. You may discover new services that your existing team is well placed to deliver.

If you are deciding where AI belongs in your marketing or business systems, speak to Net Age about the work you want to improve. Start with the business outcome, test the approach and keep people accountable for the result.

The goal is to build a stronger business. Your team should have the opportunity to help you do that.