In the AI Era, Waiting Can Be More Expensive Than Failing

Synopsis

Many business owners today are not rejecting AI, digitalisation, or technology investment. What they are really struggling with is uncertainty. They are asking whether the system they build today will still be useful two years from now, whether AI will replace the functions they are investing in, and whether a RM20,000 or RM100,000 technology project will still make sense after the next wave of innovation arrives. I understand this concern because I have faced it myself. I have invested in systems, tested ideas, and spent months developing technology that I later decided not to continue. Yet those experiences taught me something important: in the AI era, waiting for perfect certainty can become more expensive than making a controlled mistake. The answer is not to take reckless risks. The answer is to take smaller, calculated risks, learn from them, and keep moving.

Why Business Owners Are Hesitating

The hesitation I see today is not because business owners do not want to improve. In many cases, they are very aware that AI is changing the business environment. Their real concern is whether they are investing at the right time. A business owner may look at a proposal and immediately ask whether the technology will still be relevant in three years, whether something better will appear next year, or whether the entire system may need to be rebuilt after a new AI capability becomes available. These are reasonable questions because technology is evolving much faster than before. The difficulty is that there may never be a perfect moment when all the answers become clear. If a company keeps waiting for the technology to stabilise, it may eventually discover that the market has moved on while the company is still evaluating.

I Have Paid for the Wrong Technology Decisions Too

I am not writing this from the perspective of someone who always makes the right decisions. I have made expensive mistakes and invested in systems that did not continue the way I originally expected. One of the clearest examples came from an exhibition project where we needed to process a large amount of data. We needed to scrape information, organise records, analyse leads, perform outreach, and manage follow-up. Instead of depending entirely on manual work, we decided to build an application to manage the process. Programmers worked on the system for approximately twelve months, and the project gradually became more complex. If I estimate the manpower and development investment at close to RM15,000 per month, the total exposure was approaching RM180,000. After the event, however, I had to decide whether continuing to develop the same system was still the right direction.

The Hardest Part Is Knowing When to Stop

When a company has already invested a large amount of money, the natural reaction is to continue. Management starts thinking about the time spent, the development cost, the people involved, and the fact that so much work has already been completed. Walking away can feel like admitting failure. However, previous investment does not automatically justify future investment. By the time we reviewed the system, AI had advanced significantly and new approaches were becoming possible. We could see that parts of the system we had built in a traditional way might be redesigned more efficiently with AI-driven workflows. The existing system could still function, but that was no longer the most important question. The more important question was whether continuing to invest in it would still move us toward the best future position. In the end, I decided that it would not.

Not Every Failed Investment Is Completely Wasted

From a purely financial perspective, someone could say that the RM180,000 was wasted. The original system did not continue in the way we first planned, and the long-term return was not what we expected. However, from an entrepreneurial perspective, I do not see the experience as completely wasted. Because we built the system, we learned how the entire process worked in reality. We understood where scraping became difficult, where data quality broke down, where automation became unreliable, where human intervention was still necessary, and where the architecture became too rigid. We also learned what we should not build again. Without that experience, I might still be sitting here today believing that the original approach was correct. Sometimes the value of an experiment comes from proving that a direction should not continue.

Some Answers Only Appear After You Try

I have experienced the same thing with smaller experiments. At one point, I considered using AI-generated digital humans for short-form video content. I thought about it for months before finally trying it. Once I tested it, I realised that the result did not fit the way I wanted to communicate, the style of my personal brand, or the authenticity I wanted the content to have. That experiment gave me an answer that observation alone could not provide. I did not need to keep wondering whether I should use the technology. I had tested it, understood its limitations, and moved on. This is why direct experimentation matters. You can attend conferences, watch competitors, speak to consultants, and study trends, but some business answers only become clear after you use the technology in your own environment.

The Hidden Cost of Doing Nothing

Waiting feels safe because the cost appears to be zero. If a company does not spend RM20,000, the money remains in the bank. If it does not build a system, there is no development failure. If it does not test a new AI platform, there is no risk of choosing the wrong one. However, the cost of inaction is usually hidden. A company that waits for one year may also lose one year of learning, one year of process improvement, one year of staff development, and one year of understanding how AI can or cannot help the business. It may also lose the opportunity to make small mistakes while the cost of those mistakes is still manageable. The money that was not spent is easy to see, but the capability that was never developed is much harder to measure.

A Practical AI Action Plan for SMEs

For SMEs, I do not believe the starting point should be a massive AI transformation programme. The first step should be to identify one business problem that is already costing the company time, money, manpower, or missed opportunities. It may be slow quotation preparation, repetitive customer enquiries, manual reporting, lead follow-up, document processing, internal knowledge search, or data entry. Start from the business pain, not from the AI tool. Once the problem is clear, choose one small experiment that can be completed within a defined budget and time frame. An SME should be able to answer three questions before starting: what problem are we solving, how much are we willing to risk, and what evidence will tell us whether the experiment is working?

The second step is to place a ceiling on the experiment. A company that is uncertain about a RM100,000 project should not begin by signing a RM100,000 commitment. It can begin with a RM2,000, RM5,000, or RM10,000 pilot, depending on the size of the business. The experiment should be narrow enough that failure will not damage the company, but meaningful enough that management can learn something from the result. Give the test thirty, sixty, or ninety days. Measure whether it saves time, reduces manpower, generates leads, improves customer response, or produces better information. At the end of the test, management should make a deliberate decision to stop, improve, or scale.

The third step is to keep what works and remove what does not. SMEs do not have the luxury of maintaining technology simply because management is emotionally attached to it. If the experiment works, expand it gradually. If it only works partially, redesign the weak areas. If it does not work, stop and record what was learned. The objective is not to prove that every AI project succeeds. The objective is to make each experiment improve the next decision. Over time, a company builds not only technology, but internal judgment about what AI can genuinely do for the business.

Risk-Taking Should Be Controlled, Not Reckless

For me, risk-taking has never meant gambling. It means entering uncertainty with a clear limit on what I am willing to lose and a clear idea of what I want to learn. Before making an investment, I still look at the downside, question whether there is a cheaper option, and ask whether the technology is mature enough. But there comes a point when analysis must become action. If the risk is small enough and the potential learning is meaningful enough, I would rather test the idea than spend years wondering whether it might have worked. Good risk-taking also requires the discipline to stop when the evidence says the direction is wrong.

Moving Forward Without Perfect Certainty

The AI era is unlikely to give business owners the certainty they are looking for. New models will continue to appear, platforms will continue to change, and business processes will continue to evolve. The companies that perform well may not be the ones that predict every technology correctly. They may simply be the companies that are willing to test earlier, learn faster, stop bad ideas sooner, and scale good ideas with greater confidence.

My own approach has always been to move, learn, adjust, and move again. I do not believe every business needs to take a large risk, but I do believe every business should take some risk that it can afford to learn from. For SMEs, that may mean starting with one problem, one team, one process, and one manageable budget. The objective is not to become an AI company overnight. The objective is to develop the ability to experiment, make better decisions, and adapt before change becomes unavoidable.

In the AI era, the danger is not only making the wrong move. The greater danger may be spending so long waiting for the perfect move that the organisation never develops the experience required to make the next one.