Choosing the right tool for the job
LLMs are powerful but they're just one part of the AI landscape, albeit an enormous mountain
The recent surge of interest in artificial intelligence has largely focused on large language models. There’s no doubt they are impressive and have opened new opportunities for organisations of all sizes. Yet AI is a much wider field, and a focus on language models alone risks overlooking approaches that remain essential for real world results.
Traditional machine learning still delivers real value when used appropriately.
Traditional machine learning, such as predictive and classification models, continues to deliver dependable value. These techniques are well understood, efficient to run and often easier to validate and explain. In many cases, they provide better performance for tasks that involve structured data, forecasting, optimisation or decision support. They can also be deployed at lower cost and with greater transparency, which is important for organisations that operate under strong regulatory or operational constraints. If you operate in the EU, the EU AI Act is something you should know about.
Language models do have a clear place. They are powerful at handling unstructured text, supporting customer service, summarising documents and providing natural language interfaces. They can also act as reasoning engines inside agentic AI systems, which coordinate multi step tasks and automate more complex workflows. When used in the right context, these tools can significantly improve productivity, reduce manual effort and allow small teams to do more.
The future of AI belongs to organisations that focus on suitability rather than novelty.
The challenge today is that many teams feel pressure to use language models for everything. Whilst understandable, it can lead to solutions that are more expensive, harder to maintain, less reliable and slower. A short period of correction is likely as organisations refocus on choosing methods that suit the problem rather than the current trend. When this happens, AI adoption will become more stable and productive.
A balanced approach is the way forward. The most effective systems combine the strengths of traditional models and modern generative tools. For example, predictive machine learning or recommenders might drive core decision making, while a language model provides a natural language interface or helps interpret results. Agentic systems may automate workflows, but only where guardrails and clear goals are in place.
Following this principle, I help organisations adopt AI in a thoughtful, practical way through consultancy, automation, predictive modelling, recommendation systems, agentic AI and implementation. The goal is always the same: to understand the problem, examine the data and then choose the right tools. Sometimes that will be a classic machine learning model, sometimes a language model, and often a combination of both.
AI remains one of the most exciting areas of technology, and its potential continues to grow. By focusing on suitability rather than novelty, organisations can achieve reliable outcomes and build solutions that stand the test of time.
Through my consultancy services, I sit down with you and learn about your current situation and then help you navigate the options to create a plan best suited to achieving your goals.
