Artificial intelligence is becoming part of everyday business life. Whether it's helping to write an email, summarise a document, explore an idea, research a subject or tackle a problem, tools such as ChatGPT, Claude and Gemini are opening up possibilities that would have seemed futuristic only a few years ago. But there's a problem. While the technology is becoming easier to access, knowing where to start isn't always so straightforward.
Perhaps you've tried ChatGPT and received a rather generic answer. Maybe you've experimented with a few questions but haven't quite worked out how to turn the responses into something useful. Or perhaps you're interested in AI but aren't really sure what you should be asking it in the first place. You're certainly not alone.
At Fern ICT, we spend a lot of time thinking about how businesses can make practical use of AI. We talk to people who are enthusiastic about the possibilities, people who are a little apprehensive, and people who are simply wondering what all the fuss is about. One thing we have noticed is that people often need a little help getting the conversation started.
That's why we've started publishing our AI Icebreakers. A collection of practical, ready-to-use AI conversation starters designed to help you explore an idea, work through a challenge or develop something useful with an AI assistant. And the best bit? You don't need to know anything about prompt engineering to use them.
Why we’ve created the AI Icebreakers?
There’s plenty of advice available about artificial intelligence. There are courses, webinars, articles, tutorials and an ever-growing collection of people explaining how to write the perfect prompt. Some of that material is useful, but it can also make getting started seem unnecessarily complicated. We think there's a simpler way.
Most of us already know how to have a conversation. We can explain a problem, describe an idea, ask questions, challenge someone's suggestions and work towards a solution. We don't need to learn a completely new way of communicating just because the other participant happens to be an AI. Modern large language models (LLMs) are remarkably good at understanding ordinary language. You can explain what you're trying to achieve, provide some background and ask for help. You can then respond to what comes back, correct misunderstandings and develop the conversation. The difficulty is often knowing what to ask first, particularly when you're exploring something unfamiliar.
Our Icebreakers are intended to bridge the gap. Each one provides a starting point, introduces a little structure and encourages the AI to guide you through a conversation rather than simply produce a one-off answer. They are not intended to do all the thinking for you. Quite the opposite. They are designed to get you thinking, asking questions and exploring possibilities that you might not otherwise have considered.
We also want people to experiment. You don't need a business case, a formal AI strategy or an expensive subscription to start discovering what these tools can do. A free account with one of the mainstream AI services is enough to begin exploring many of the possibilities.
A closer look
Let's take one of our Icebreakers and look at how you might use it. The Explore a Business Idea conversation starter is designed for anyone who has an idea they'd like to investigate. It might be a completely new business, a product or service, an opportunity within an existing organisation, or simply a problem that you think somebody might be willing to pay to solve. You don't need a polished business plan or even a particularly well-developed idea. In fact, a rough idea is often a good place to start.
The conversation starter asks the AI to act as an experienced business adviser, helping you explore the idea through a series of questions and discussions. Rather than immediately producing a business plan, it encourages the AI to understand what you're thinking, investigate the problem and gradually develop a clearer picture.
Imagine you've been thinking about providing a new service to local businesses. You have a vague idea of what it might involve, but you're not sure who would buy it, what they would pay or whether there's really a market. You copy the conversation starter into ChatGPT, Claude or Gemini and begin.
The AI starts by asking you to describe your idea. You explain it in your own words. It might ask who you think the potential customers are, what problem you're trying to solve and how people currently deal with that problem. You answer, perhaps adding details you hadn't originally considered. The AI asks further questions and starts identifying some of the assumptions behind your idea.
At this point, you might begin to see the real value of the exercise. The AI isn't necessarily telling you anything revolutionary. It is helping you organise your thinking, examine your assumptions and look at the idea from different perspectives. You might discover that your original idea is too broad, that there is a more interesting customer problem to address, or that you have overlooked an existing competitor. Equally, you might uncover an opportunity worth investigating further.
As the conversation develops, you can ask the AI to produce a summary of the opportunity, identify the biggest risks, suggest ways of testing customer interest or develop a simple plan for validating the idea. You could even ask it to prepare a short business proposition that you can share with someone else for feedback. The important thing is that the outcome develops through the conversation. You are not simply accepting a business plan generated from a few sentences of input. You are contributing your knowledge, experience and judgement throughout the process.
That distinction matters. The AI can help you explore an idea, but it cannot establish whether customers will actually buy your product or service. That requires evidence, conversations with potential customers and, ultimately, testing in the real world. The Icebreaker gives you somewhere useful to start.
Getting more from your AI conversation
One of the most common mistakes people make when using AI is treating the first response as the finished answer. It rarely needs to be. Think of the conversation as working with a colleague. You wouldn't expect a colleague to understand everything about a task from one sentence, nor would you necessarily accept their first suggestion without discussion. The same approach works remarkably well with an AI assistant. The more useful context you provide, the better the AI can understand what you're trying to achieve. Tell it about your organisation, your customers, the problem you're facing, what you've already tried and any constraints you need to work within. You don't need to provide everything at once. Add information as the conversation progresses.
If you're not sure what information is missing, ask the AI. Something as simple as "What else would you need to know to give me a more useful answer?" can be surprisingly effective. You can also be much more specific about the sort of response you want. Perhaps you need a short executive summary rather than a lengthy explanation. Maybe you want a formal report, a business proposal, a presentation outline, a table comparing options or a set of practical actions. Tell the AI what you intend to do with the output and who will be reading it.
If the response isn't quite right, say so. Ask it to change the tone, simplify the language, provide more detail or approach the problem from another angle. You can ask for examples, challenge an assumption or tell it to reconsider a particular recommendation. This back-and-forth dialogue is where much of the value lies. Don't accept the first response as final. Treat it as the beginning of a discussion.
Try changing the AI's role
Another useful technique is to ask the AI to look at the problem from a different perspective. For example, when exploring a business idea, you might initially ask it to act as an experienced business adviser. Later, you could say: "Now I'd like you to review everything we've discussed from the perspective of a business coach or mentor. Be constructive but challenge my thinking. What am I overlooking? Where might I be making assumptions without sufficient evidence?"
You could ask it to take on the perspective of a potential customer, a marketing specialist, a financial adviser or even a sceptical investor. These are not necessarily different AI systems or specialist products. You're simply asking the model to consider the same information through a different professional lens.
You can also ask it to review its own recommendations. For example: "Please cross-check your conclusions. Identify anything that might be inaccurate, unsupported or based on an assumption. Explain why you reached your conclusions and tell me where further evidence is needed."
It's worth remembering that an AI's explanation isn't proof that its answer is correct. Nor should you rely on it to provide a complete record of its internal reasoning. What you want is a clear explanation of its conclusions, assumptions, supporting evidence and any uncertainties. For factual work, ask it to identify sources so that you can check them independently.
Don't be afraid to tell it when it's getting things wrong!
Sometimes an AI conversation can become a little dull. Responses might become repetitive, overly enthusiastic or full of generic suggestions. Occasionally, the model seems to lose sight of what you're actually trying to achieve. Don't put up with it! Tell it that the responses are becoming repetitive, that it is missing the point or that you want it to stop making the same suggestions. Ask it to be more direct, more critical or more creative.
You can even say: "You're going round in circles a little. Let's stop repeating the same points and look for some genuinely different approaches." Or: "I'm not finding these suggestions particularly useful. Please rethink your approach and ask me some better questions."
The AI won't be offended. It doesn't have feelings to hurt, and you don't need to worry about being impolite. Clear feedback helps it adjust its responses to what you actually need.
You don't have to type everything
Although our Icebreakers are supplied as text that you can copy into a conversation, that doesn't mean you need to spend the next hour typing. Many AI services now offer voice conversations. Once you've started an Icebreaker and established some context, you can switch to voice mode (where supported) and continue talking naturally.
This can be particularly useful when you're exploring an idea, thinking aloud or working through a problem that is difficult to explain in a few carefully written sentences. You might find that speaking is actually a much more natural way to work. You can explain something, hear the response, interrupt, ask a follow-up question and explore a different direction without constantly switching between thinking and typing.
You can also use voice to ask the AI to summarise where you've got to, identify outstanding questions or suggest what you should explore next. It can feel much more like a conversation with a colleague than using a traditional computer application.
Try the same Icebreaker with different AI models
Here's another experiment we'd strongly encourage. Take exactly the same conversation starter and try it with different AI services. For example, you could use ChatGPT from OpenAI, Claude from Anthropic and Gemini from Google. You may be surprised by the differences.
Each model has its own characteristics. Some may ask more questions before offering suggestions. Others might be more direct, more analytical or more creative. They may interpret your answers differently, identify different opportunities or approach the problem from an entirely different perspective. There isn't necessarily one correct response.
Trying the same Icebreaker with different models is a useful way to discover which ones you enjoy working with and which are better suited to particular tasks. It also helps you avoid becoming overly dependent on a single AI service. You might even take an interesting observation from one conversation and introduce it into another. For example, you could ask a second model to review the conclusions reached by the first and identify anything it agrees or disagrees with.
Just remember that agreement between two AI models doesn't automatically make something true. They can share the same assumptions or make similar mistakes. Use the comparison to broaden your thinking, not as a substitute for checking important information.
Is prompt engineering still important?
If you've spent any time reading about AI, you'll probably have come across the term prompt engineering. It describes the process of designing and refining instructions to get better results from an AI model. There is a whole discipline around it, particularly when building AI applications, automating processes or developing complex systems.
But does the average person need to learn prompt engineering before they can use AI effectively? Our view is that the need to become a specialist prompt engineer is diminishing for everyday users. The latest generation of large language models is increasingly capable of understanding ordinary language, interpreting context and working with people through an extended conversation. You no longer need to spend hours learning complicated prompt structures or trying to discover a particular combination of words that will make the AI respond correctly.
That doesn't mean the way you ask questions is irrelevant. Being clear about what you want, providing useful background information and explaining what a successful outcome looks like will always help. It simply means that you can increasingly develop the conversation naturally rather than trying to construct one perfect instruction at the outset. Some of the most useful questions you can ask an AI are about the conversation itself:
"Have I explained that clearly?"
"What am I missing?"
"What questions should I be asking?"
"How could we improve this approach?"
"What would you recommend we do next?"
The AI can help you formulate better questions, identify gaps in your thinking and suggest how to approach the task. In effect, it can help you become better at working with AI while you're actually using it. As models become more capable, the emphasis is likely to move further away from clever prompt construction and towards providing the right context, setting objectives, exercising judgement and working collaboratively with AI. That is a much more accessible prospect for most people and businesses.
AI adoption. Where should you start?
There’s a lot of discussion about AI adoption, digital transformation and the impact artificial intelligence will have on businesses. It can sometimes feel as though everyone else has already worked out what to do. The reality is rather different. The UK Government's 2026 AI Adoption Survey found that around one-third of UK businesses were using at least one AI technology, up from 16% in 2025. It also highlighted continuing barriers around skills, expertise and understanding how AI can be used. Concerns about accuracy, bias, privacy and security remain important too. In other words, adoption is growing quickly, but many organisations are still working out where AI fits and how to use it effectively.
We think the best place to start is with something relatively simple. Don't begin by asking how you can introduce AI across your entire organisation. Start by looking at the things you already do. Is there a task that takes too long? Do you regularly prepare similar documents? Is there research or analysis that could be made easier? Are there ideas or opportunities you haven't had the time to explore?
Pick something that matters but isn't critical if the first attempt doesn't work particularly well. Try using an AI assistant to help with it. Experiment, see what happens and consider whether the result is genuinely useful. If you're a business owner, you might start by asking an AI to review a service description, help prepare a customer communication, explore a marketing idea or identify potential improvements to an internal process.
If you're an individual, perhaps you want to learn about a new subject, plan a project, organise your thoughts or develop an idea. The important thing is to start with a real task or question rather than adopting AI simply because it seems to be what everyone else is doing. As you become more familiar with the technology, you can begin to explore more substantial opportunities. You might discover ways to improve workflows, support employees, develop new services or make better use of information already available within your organisation.
Of course, businesses also need to think about the information they share with AI services, data protection, confidentiality, security and the reliability of the outputs. Experimenting is a good way to learn, but it should be done with an appropriate understanding of the risks. Avoid putting confidential, personal or commercially sensitive information into a service unless you understand and have approved the relevant arrangements. The objective isn't to introduce AI everywhere. It's to discover where it can make a worthwhile difference.
The Icebreakers are just the beginning
We want to be clear about something. Our AI Icebreakers are starting points, not magic formulas. There are no guarantees that every conversation will be successful or that the AI will always understand what you're trying to achieve. You might get an excellent result straight away. Equally, you might need to explain things differently, correct misunderstandings or take the conversation in another direction. Sometimes you may decide that a particular approach isn't working and start again. That's perfectly fine.
The real value comes from keeping the conversation going, providing guidance and working with the AI to develop the result. Think of the Icebreaker as opening the door. What happens next is up to you.
We also want to hear from people who try them. Did the conversation starter help you get going? Did it ask useful questions? Was the outcome what you expected? Was there something you found confusing or frustrating? Perhaps you discovered a better way of using it. Your feedback will help us improve the existing Icebreakers and develop new ones. You can find the current collection here: https://fern-ict.co.uk/icebreakers/. They're free for you to use, experiment with and adapt to your own circumstances.
Ready to give AI a try?
Whether you're completely new to AI, have experimented with an LLM a few times or are already using AI regularly in your work, we'd encourage you to jump in and explore. Try one of the Icebreakers. Use it with different AI models. Ask questions, challenge the answers and see where the conversation takes you. If you're not sure where to start, that's fine too. At Fern ICT, we're keen to help people and organisations at every stage of their AI journey. Whether you need a little guidance getting started, want to understand the opportunities for your business, or are considering how to develop a more structured approach to AI adoption, we're always happy to have a conversation.
Have a go, see what you discover and let us know how you get on.
Explore the Fern ICT AI Icebreakers or get in touch with us. We'd particularly welcome your feedback on the conversation starters and ideas for what we should create next.
Fern ICT Limited – helping people and organisations make practical, informed use of artificial intelligence.