Why We Rebuilt McCaigs Around Deterministic Systems Instead of AI Hype

Over the last year, something unexpected happened. The more AI systems we built, the less convinced we became that AI was the answer to every problem.

That might sound strange coming from a technical studio that spends its days building intelligent products, automation systems and digital platforms. But real-world engineering has a habit of exposing reality very quickly. When you work with actual businesses rather than technology demonstrations, you discover that most problems are not caused by a lack of artificial intelligence. They are caused by a lack of systems.

Poor information flow, disconnected processes, knowledge trapped inside people's heads, manual administration, slow response times, missed enquiries and inconsistent customer experiences are ordinary business problems. They are also expensive ones. The more projects we worked on, the more obvious it became that many businesses were being sold AI when what they actually needed was something simpler: better engineering.

That realisation led us to redesign both our services and our positioning. Today, McCaigs is positioned as Scotland's Elite Technical Studio: a hands-on studio focused on solving practical business problems through intelligent engineering.

The problem with starting with AI

AI is extraordinary. We use it every day, build with it and experiment with it constantly. But there is a growing assumption that every business problem needs a large language model attached to it. In reality, that often introduces unnecessary complexity.

Imagine a garage owner receiving twenty phone calls every day. Customers ask whether they can get a quote, whether the garage services a particular vehicle, how long a repair might take, whether the business is open on Saturdays and what happens next. The same pattern exists in cafés, estate agencies, trades businesses, professional services firms and local companies of every kind.

Many of those businesses are immediately told that they need an AI chatbot. Often they do not. They need a system that understands a limited set of business questions and returns accurate, reliable answers every time. The challenge is not intelligence. The challenge is organisation.

Don't pay for intelligence when logic is enough.

What building real systems taught us

As we developed internal tools, client projects and our own software products, a pattern emerged. Most business workflows are surprisingly predictable. Customers ask similar questions. Staff follow similar procedures. Leads move through similar stages. Information follows repeatable paths.

The more predictable a process becomes, the less AI is required. Instead of generating answers, systems can retrieve approved answers. Instead of guessing, they can know. Instead of attempting to interpret everything, they can guide people through structured journeys.

This creates a different approach to software design: one that is often faster, cheaper and more reliable. AI still has a valuable role, but it is used where judgement, language, classification, summarisation or ambiguity genuinely call for it.

Practical intelligence for ordinary businesses

The market often talks about AI as if the only worthwhile opportunities belong to large enterprises. That overlooks the reality of the UK economy. SMEs, sole traders and owner-managed businesses have practical problems worth solving properly.

An estate agent may need a system that answers questions about available properties, captures buyer requirements and schedules viewings. A garage may need to collect vehicle details and prepare useful diagnostic information before a technician becomes involved. A trades business may need to gather project information, request photographs, route urgent work and book surveys. A café or hospitality business may need clearer enquiry handling, reliable booking information and less repeated administration.

None of these systems needs to be theatrical. The value comes from making everyday work easier, faster and more consistent.

The McCaigs approach

Our own deterministic Assistant demonstrates the principle. At first glance, it feels like an AI product. Responses appear progressively, the interface is conversational and the experience is modern. Underneath, approved knowledge, controlled workflows, context-aware routing and business rules do most of the work.

That changes the economics of the system. Responses can be fast. Outcomes can be predictable. Answers can be traced. Workflows can be reviewed. Where a question falls outside the approved knowledge, the system can say so clearly and route the person towards the right next step.

For many businesses, that is more useful than a general-purpose assistant trying to sound confident about everything.

Why we repositioned McCaigs

As our thinking evolved, our original positioning no longer reflected the work. The market was becoming crowded with AI agencies. Every company claimed to be revolutionary. Every landing page promised transformation. Every product claimed to change everything.

Meanwhile, ordinary businesses still had ordinary problems: missed leads, manual processes, slow administration, disconnected systems and customers waiting for answers. Those businesses do not need hype. They need practical outcomes.

McCaigs was rebuilt around that reality. We diagnose the awkward part of the operation, design the simplest reliable route forward and build the system with modern tools. Sometimes AI is part of the answer. Sometimes automation, a rules engine, a website, an internal tool or a better information flow is the right answer.

The outcome matters more than the label.

Building useful systems quickly

Modern platforms allow a small technical studio to move quickly without treating speed as an excuse for poor engineering. Proven infrastructure, careful interfaces and a narrow first release make it possible to build useful systems in weeks rather than allowing a project to drift for months.

That matters for SMEs because time, budget and attention are finite. A focused first release can improve one real workflow, create evidence and show what deserves to happen next.

The businesses we work with care less about which model powers a system than whether it saves time, improves the customer experience or helps the team work more clearly. They are right to think that way.

Knowing when to use each

The future is not AI versus deterministic systems. The future is knowing when to use each.

If a workflow is predictable, repeatable and structured, deterministic software is usually the better foundation. If a workflow requires creativity, interpretation, reasoning, summarisation or careful handling of ambiguity, AI may be the right tool. Often the most effective system combines both.

The important thing is to begin with the business problem rather than the fashionable technology. In many cases, the smartest system is the one that does not need AI at all.