How to get the best results from GPT5

Why GPT-5 requires a different approach

If you’ve tried GPT-5, you might have noticed it doesn’t always give you the answer you hoped for. That’s because GPT-5 is very powerful, but it expects you to be clear and specific in what you ask. With a few simple techniques, you can dramatically improve the quality and speed of its replies.

Prompting tips for GPT-5

With GPT-4, you could often type in a rough request and get a decent answer. GPT-5 is more like working with an expert, it can do incredible work, but only if you brief it properly. If your instructions are vague or contradictory, it will struggle. If your instructions are clear, you’ll see far better results.

Make instructions explicit  

Paste in a short piece of text you like and ask it to copy the tone or structure

Break big tasks into steps

Tell it if you want the answer to be professional, casual, short, or detailed

Ask it to check its own work

One of the most effective strategies is to ask GPT-5 to critique your prompts. Provide an example, specify the behaviour you want versus what you’re getting, and request minimal adjustments. This turns GPT-5 into its own prompt coach, rapidly iterating toward stronger performance.

Speeding up responses

Sometimes GPT-5 takes a while to respond, especially with complex requests. You can prompt it to speed up the response:

Ask it to be brief: “Answer in 5 bullet points"

Use simple, direct instructions instead of long paragraphs

If you don’t need deep reasoning, tell it: “Be quick and focus on practical tips”

Why this matters for businesses using AI

Think of GPT-5 as an employee. If you give it a sloppy brief, you’ll get sloppy work. If you give it a clear, detailed brief, you’ll get high-quality results. With practice, you’ll be able to turn GPT-5 into a reliable assistant that saves you hours every week

For small and medium businesses this can be the difference between clunky automation and incredible productivity. The Breezy AI booking assistant is built on top of large language models, which means prompting best practices are core to how we deliver reliable, human-like support. If you’re curious how this works in practice, read our post on AI Chatbots vs AI Agents