New developments in AI everyday, and GPT-5 is also here. GPT-5 while might have faced certain backlash for certain missteps, yet it is leap forward because there’s no doubt that there have been major advancements. Because when compared to GPT-4 it’s faster, more consistent, and significantly better at reasoning through complex instructions. But there’s a small catch, GPT-5 is only as powerful as the way you use it. As good is your input that good will your output from the AI model. Therefore, if you give a vague prompt, the chances of getting a vague prompts. Prompt engineering is everything, a structured, specific prompt, can transform GPT-5 into a sharp productivity partner. After a lot of trial and error, these are the five prompts that now save me the most time, paired with notes on what didn’t work and why.
The simple formula for better GPT-5 prompts
You don’t require the use of complicated techniques of prompt engineering for the attainment of better output from GPT-5. Most of the time, all that is required is to provide the model with a suitable context that would help it understand what it is expected to do.
A useful prompt consists of the following components:
- Role: Specify for GPT-5 the persona it is expected to adopt. This could be an editor, planner, researcher, or an analyst, for example.
- Task: Indicate clearly what you want GPT-5 to do.
- Context: Provide relevant information which is necessary to achieve that goal.
- Constraints: Give instructions about format, length, audience, or tone.
1. The weekly planner prompt
Prompt I use:
“Act as my productivity coach. Plan my week in blocks of deep work, meetings, and recovery. Keep my mornings for creative work, and limit calls to afternoons.”
What works: GPT-5 works and understands constraints really well. So when prompted something like “mornings = creative work”, it took that into consideration and structured my calendar accordingly.
What didn’t work: Giving it generic prompts such as asking it to “just make me more productive”. Because such basic prompts gave me the most basic answers like “wake up early” or “avoid distractions.” Which I know and I don’t want to hear it again from an AI Model.
Impact: This prompt has helped me saved 2 hours from my week and has also reduced and decision fatigue.
2. The instant briefing prompt
Prompt I use:
“Summarize the top 5 news stories in [industry]. Give me headlines + 2-sentence context + why it matters.”
What works: Adding “why it matters” turns a bland summary into something insightful, almost like a mini-briefing from an analyst.
What didn’t work: When I didn’t give the model any instruction, I received dry summaries which I could have found through a quick Google search.
Impact: The effect of a prompt like this is that it has made my daily news scan from 20 minutes to 3 minutes.
3. The brainstorm and refine prompt
Prompt I use:
“Give me 10 unusual ideas for [task]. Then refine the best 3 into actionable steps I can try this week.”
What works: GPT-5 has proven to excel when it is asked to generate and then filter. This helps as it avoids overwhelm and delivers a clear shortlist.
What didn’t work: Previously, and up until a while back I used to ask “ give me 10 ideas” without refinement, which left me with an unfocused list.
4. The draft-polish prompt
Prompt I use:
“Here’s a rough draft of my email/blog/post. Rewrite it to sound professional yet friendly, keep it under 200 words, and remove jargon.”
What works: GPT-5 is actually much better at tone control than previous versions. The rewrites are polished but most importantly don’t sound robotic.
What didn’t work: Asking it to “make this better”, this often led to unnecessarily flowery language and longer sentences, and it almost always sounded robotic and AI generated.
Impact: This cut editing time by more than half. This task which used to previously take me 20-minutes, now takes less than 10.
5. The ‘Why am I stuck?’ prompt
Prompt I use:
“I’m stuck on [task]. Ask me 3 clarifying questions, then suggest the next step I should take.”
What works: Instead of giving a rushed solution, GPT-5 asks targeted questions, which often reveal gaps in my thinking.
What didn’t work: At first when I used to ask it to “solve the problem” directly, it actually gave me vague and generic answers without any context.
Impact: This helped me move past roadblocks in minutes, when before I used to waste more than half an hour staring at the task.
Treat prompting as an iterative process
One of the errors I made when I first began using GPT-5 was assuming the first prompt would yield a definitive answer.
This is seldom the case.
Even an excellently phrased prompt will yield an answer that may be accurate but fail to provide the desired tone, depth, or format. As opposed to starting again, I found it better to see the first input as something I can build upon.
For example, suppose GPT-5 gives you a lengthy and general answer. You may then follow up with:
“Make the above response specific to SaaS founders. Remove generalities and include three specific examples.”
If the tone sounds formal:
“Rephrase the above in a conversational tone while maintaining the ideas.”
Or if there is too much information:
“Reduce this by 40% to include only the most relevant points.”
The main thing is that conversation should be developed rather than aiming to create one perfect prompt. Often, the best outcomes come from a specific cycle – prompt, receive, refine, repeat.
The Big takeaway
The experiments that failed shared one thing in common: they were too broad, too generic. Generic prompts produced generic output. The prompts that have proven to work are those which are specific, clear and well drafted. So, if you want GPT-5 to actually transform your workflow, start by refining your prompts, use the prompts I’ve listed above and see. Test them first, notice what falls flat, and adjust until the AI becomes less of a chatbot and more of a partner in how you think and work. If you want to go a step further and explore how to supercharge your results by combining ChatGPT (or GPT-5) with other AI tools, check out this practical guide on how to integrate multiple AI tools for stronger outcomes.
