Prompt Engineering
Why Good Communication Skills Matter More Than Ever
Artificial intelligence is only as useful as the instructions we give it. That’s why prompt engineering – the craft of structuring clear, purposeful inputs – has become a decisive factor in whether AI feels like a powerful partner or a frustrating experiment. It’s becoming a new skill in its own right and we have started to see it being listed as a requirement on job specs in the AI sector.
In essence, this is really just about having good communication skills, something which the cultural sector already excels in. The only difference is that you have to align yourself with the LLM rather than another person to get the outcome you want.
For workers in the arts and cultural sector, where staff time is precious and customer expectations are high, it pays to learn how to do this effectively. Not doing so can lead to spending hours playing with an LLM and only getting sub-par, generic results that you have to spend more yet more time refining before it’s any good. Trust us, we’ve learned the hard way!
Why Prompts Matter
AI models don’t “think” like humans. They respond to patterns in language. They’re essentially a very very large prediction engine. They output the words (technically, they actually output just the characters i.e. individual letters), that are most likely to come next (based on their training data), given the previous request. A well-crafted prompt guides the system to focus on the right information to give the best answer, not just spit out generic, general purpose text.
Consider these two versions of the same instruction:
Loose prompt: “Tell me about Romeo and Juliet.”
Refined prompt: “Give me a 200 word summary of Shakespeare’s Romeo and Juliet written for a theatre audience, highlighting the themes of love and conflict.”
The difference is stark. The first could return anything from a literary analysis to a general Wikipedia-esque overview. The second produces something tailored, concise, and directly useful.
It does this by specifying:
- The length of output wanted
- The topic, with specificity i.e. R&J “highlighting the themes of love and conflict”
- The intended reader for the 200 word summary
Here’s an even more detailed one:
Detailed prompt: “Generate a 400 word summary of Shakespeare’s Romeo and Juliet for our theatre programme. The audience will be a mixture of family and friends for a schools performance. The performers are aged 14 – 18 and have been rehearsing this for only a week in our intensive summer camp. Emphasise the themes of love and conflict and how our production relates this to the modern day themes.”
Precision is key. You can see how, regardless of these instructions being directed to a human or an LLM, the potential outputs are vastly different.
The Benefits of Clear Prompting
1. Consistency – Well-designed prompts keep LLMs on topic and stop them drifting into irrelevant tangents or “hallucinating” false or inaccurate information.
2. Efficiency – Clear prompts reduce wasted queries and speed up useful outputs.
3. Tone and Audience Fit – You can specify whether an answer should be professional, warm, or explanatory and tailor it to different reading ages or audience segments.
4. Bias Control – Prompts help constrain AI to fact-based, accessible answers rather than speculation.
In short, prompting transforms AI from a generalist tool into a sector-specific expert assistant.
Examples in Practice
Here are a few more example prompts:
Social media
Loose: “Write some social media posts to promote our panto.”
Refined: “Write 10 social media posts for Twiiter/X promoting our pantomime targeted at families with young children. Highlight our family tickets which include an ice cream.”
Marketing
Loose: “Write something engaging about our new exhibition.”
Refined: “Draft a 500-word blog post about our new exhibition on modern sculpture. Emphasise sustainability themes and include a clear call to action to book.”
Customer accessibility queries
Loose: “What access support is available?”
Refined: “Explain the access provisions for customers attending performances at our main theatre. Keep the tone welcoming and concise, broken into bullet points.”
These examples show that prompts act as both the brief and the quality control.
A Good Shortcut
If you’re really short on time or don’t quite know what you need but have a good idea, here’s a cheat code: ask another AI model to improve your prompt for you!
Example: “Improve this prompt: Rewrite our terms and conditions as an easy reader”
Improved Prompt Output (using Google’s Gemini 2.5 Flash model):
“Please rewrite our company’s terms and conditions to be easily understandable for the average user.
Your goal is to transform the legal jargon into plain, simple language. The rewritten text should have a sixth-grade reading level.
Here are the key requirements:
- Avoid legal jargon. Replace complex terms with simple, everyday words. For example, change “indemnify” to “protect us from legal claims.”
- Use a conversational and direct tone. Use “you” and “we” to make the text feel more personal and less formal.
- Keep sentences short. Break up long sentences and paragraphs into smaller, more manageable parts.
- Highlight key information. Use bolding, bullet points, and headings to make important points stand out.
To give you a better idea of what we’re looking for, here’s an example of a good rewrite:
Original Legal Text: By accessing and using this service, you agree to indemnify and hold harmless the company from any and all claims, damages, expenses, or liabilities arising from your use.
Easy Reader Rewrite: When you use our service, you agree to be responsible for your actions. This means you will protect us from any legal issues or costs that might come from something you do.”
Now you can use this prompt, upload your real terms and conditions and be much more confident the output will be usable.
Why You Should Care
Hopefully this shows that AI isn’t as simple as plug-and-play. AI needs carefully designed prompts to perform to a set standard. Prompt engineering is about clarity, audience awareness, and intentionality – principles the arts sector already understands well.
Done right, this doesn’t just save staff time. It builds trust in AI usage by both internal staff and customers, boosts operational efficiency, strengthens accessibility, information architecture and ensures every interaction reflects the values of your organisation and conforms to a set standard.
At SynapTix, we see prompt engineering as one of the most important crafts that makes our agent products effective. With the right words, AI can be a powerful extension of your team, a domain expert not an intern. If you need a head start, you can head to our prompt library, specifically tailored to common tasks within the arts and cultural sector and even contribute one yourself!