AI may feel instant and invisible, but every prompt carries a cost. Sarah Bailey, Head of Marketing at Nexus Open Systems, explores why more thoughtful, purposeful use of AI can deliver better results for businesses while reducing wasted time, resources and effort.

Artificial Intelligence has quickly become part of everyday business life. Whether we’re summarising meetings, drafting emails, analysing information or researching new ideas, AI tools are helping people work faster than ever before. As adoption grows, most conversations focus on productivity, efficiency and innovation.

Far fewer focus on what sits behind the prompt.

Every AI interaction relies on a vast network of infrastructure: data centres, servers, cooling systems, specialist chips and power networks. AI may feel instant and invisible from a user perspective, but delivering those answers requires real-world resources.

As demand for AI continues to grow, technology providers are investing heavily to ensure that growth happens in a more sustainable way. Microsoft, for example, has introduced new datacentre designs that use closed-loop cooling systems with zero water evaporation for cooling and continues to work towards ambitious goals around carbon, water and waste reduction. Microsoft has also reported significant improvements in datacentre water efficiency compared to its earliest facilities.

That is encouraging progress, but sustainable AI is not just a challenge for infrastructure providers.

It’s a challenge for all of us.

The good news is that sustainable AI is not about using less AI. It is about using AI more intelligently.

For many organisations, the biggest cost of AI is not the electricity required to generate an answer. It is the human time that follows. A vague prompt often creates a vague response. That response then needs reviewing, correcting, refining and sometimes repeating entirely. The environmental impact may be difficult to see, but the operational impact is obvious.

Multiply that process across an organisation, with dozens or even hundreds of employees using AI tools every day, and small inefficiencies can quickly add up. Poorly defined requests do not just waste computing resources. They can create additional work for the people using the technology and undermine the productivity gains that AI was introduced to deliver in the first place.

Think about the difference between these two requests:

“Summarise this 200-page document.”

and

“Create a chronology of key events from pages 44-68 and provide a page reference for each point.”

Both use AI.

One is likely to produce a broad, generic output requiring significant review. The other has a clearly defined purpose, a limited source and built-in traceability. It is likely to be faster to verify and more useful to the person receiving it.

The lesson is simple: better prompts create better outcomes.

This does not mean every employee needs to become an expert in prompt engineering. In many cases, it simply means being clear about what you need before asking AI to produce it. Providing context, defining the required format and narrowing the scope of a task can make a significant difference to the usefulness of the response.

For businesses across the South West and beyond, that means viewing AI through a slightly different lens. Rather than asking, “Can AI do this?”, we should be asking:

• Is AI the right tool for this task?
• What is the smallest useful piece of work I need it to perform?
• How will I verify the result?

These questions improve quality, reduce wasted effort and help ensure technology is being used where it adds genuine value.

They also encourage businesses to think more carefully about where AI fits within their wider processes. Automating or accelerating a task does not automatically make the overall process better. If the output still requires extensive checking, rewriting or manual intervention, organisations need to consider whether AI is genuinely improving the workflow or simply moving the work elsewhere.

This idea aligns closely with something we often discuss with clients at Nexus. Technology should never be adopted simply because it is new. It should be adopted because it solves a problem, improves a process or creates a measurable benefit. In a recent article, I wrote that competitive advantage will not come from adopting AI the fastest. It will come from adopting it wisely.

That feels even more relevant today.

AI is becoming business infrastructure in the same way that cloud computing did before it. As it becomes embedded in more everyday tools and processes, organisations will need to move beyond experimentation and think about how it is being used across the business.

That means considering not only what AI can do, but where it adds value, where human judgement remains essential and how employees can use these tools effectively. Clear policies and training have an important role to play, but so does something much simpler: encouraging people to approach AI with purpose.

The organisations that succeed will not necessarily be those generating the highest volume of AI content. They will be the ones using AI deliberately, efficiently and responsibly.

Ultimately, the most sustainable prompt is not the shortest prompt.

It is the prompt that creates the greatest value with the least waste.

As AI continues to evolve, perhaps the most important habit we can develop is simply pausing before we type and asking one question:

What outcome am I actually trying to achieve?

Because when we think before we prompt, we usually get better answers, better results and better value from the technology we’re using.

Sarah Bailey, Head of Marketing, Nexus Open Systems