Artificial Intelligence is undoubtedly the buzzword of the decade – and as it dominates boardroom agendas and LinkedIn feeds, Kevin Abela, Managing Director of leading cloud services provider Tristratus, believes that distinguishing between marketing hype and operational reality has never been more critical.

“The first thing we do is strip away the hype,” Kevin states matter-of-factly, discussing how Tristratus helps clients identify the practical reality of what these tools can actually achieve. “A lot of what is being marketed as AI is either standard automation with a new label, or a demo that looks impressive but has no measurable business value.”

For a Maltese SME, the litmus test for any new tool shouldn't be whether it boasts an AI label. Instead, the Managing Director insists, leaders must ask if it saves time, reduces errors, improves response times, or increases conversion. “If it does not move one of those numbers, then it is probably hype.”

There is a palpable fear of missing out among business leaders today, driving many to invest in AI simply to avoid being left behind. However, buying sophisticated software without a solid foundation is akin to buying a Ferrari just to drive to the corner shop.

Kevin identifies a significant gap between what a CEO expects AI to achieve and the reality of its implementation. “The biggest disconnect is that many business leaders think AI is a turnkey replacement for judgment, process, and discipline. It is not,” he warns.

While AI can accelerate work through pattern recognition, summarisation, and drafting, it cannot act as a plaster for deep-rooted operational flaws. “It does not fix broken processes, poor data, unclear ownership, or weak decision-making. If those issues exist, AI usually makes the mess faster, not better.”

He also notes that businesses routinely underestimate the foundational work required to make AI functional. “Value comes from defining the use case properly, cleaning data, setting guardrails, assigning ownership, and measuring outcomes. If you are spending money on AI while your staff are still passing critical information through WhatsApp, scattered spreadsheets, and disconnected inboxes, you are solving the wrong problem in the wrong order.”

Rather than leaping straight into the glamorous world of generative AI, Kevin champions a more pragmatic first step.

“Boring automation pays,” he explains. “Generative AI gets attention, but structured process automation usually delivers value faster, with lower risk and clearer ROI.”

In the local landscape, the most successful technological deployments are rarely customer-facing chatbots. Instead, they operate quietly behind the scenes. Kevin points to tangible, immediate wins: automatically routing enquiries, generating tasks from forms, moving approved sales into delivery workflows, and eliminating double entry between systems.

“In many SMEs, the first win is not content generation. It is removing manual steps that never should have existed in the first place,” he maintains. When combined with AI – for instance, extracting key details from an incoming email and automatically routing it to the correct team – the bottom-line effect becomes highly tangible. “It is reducing cycle time and improving execution.”

Malta’s notoriously tight labour market has left many teams stretched thin. Consequently, there is an underlying workforce anxiety regarding AI obsolescence. However, Kevin believes leaders should frame the conversation around enablement, rather than replacement.

“For most Maltese SMEs, AI is more useful as a force multiplier than as a replacement strategy,” he argues. “I do not see the main opportunity as ‘replace people.’ I see it as ‘make good people more effective.’”

By allowing employees to handle more volume and respond faster, AI empowers smaller local teams to operate with the consistency and scale necessary to compete with larger international players. As for the workforce, Kevin's advice to management is clear: “Most businesses do not need ‘AI-whisperers.’ They need employees who understand their job well, can think critically, and can learn how to use new tools responsibly.”

Innovation, however, cannot come at the expense of governance. The horror stories of rogue chatbots offering unauthorised discounts or hallucinating facts serve as a cautionary tale.

To this end, the Managing Director outlines strict red flags for business leaders: “Any AI tool that can communicate externally without guardrails should be treated as a risk. If it can respond to customers, generate commitments, or provide advice without approval logic, that is a control issue.” Furthermore, he stresses that AI must be restricted to approved data sources to prevent hallucinations, and a human process owner must always remain accountable for the output.

When it comes to the EU’s stringent data privacy laws, Kevin is equally uncompromising. “You do not stay compliant by avoiding innovation. You stay compliant by putting controls first,” he states. This necessitates rigorous data classification, approved-use policies, and vendor due diligence. If a company's data is currently fragmented across scattered Excel sheets, achieving data readiness must become the first project, not the side issue.

Fast forward three to five years, and Kevin predicts the current AI frenzy will have fundamentally shifted. “I think the label will become less important. AI will stop being treated as a category and will become part of normal business systems, just like cloud, mobile, or workflow software did.”

A mature AI environment will not be defined by companies chasing the latest standalone tool, but by technologies deeply embedded into CRM, finance workflows, and compliance processes.

Finally, I ask, what is the single most important thing a business owner should do before purchasing a software licence or hiring a consultant?

“Define the business problem in measurable terms, not just ‘we want to use AI’,” Kevin concludes. “Define what is actually broken. What is too slow, too manual, too costly, or too inconsistent? What would improvement look like in numbers? If that is not clear, then buying software is just guesswork.”