Turn prompting into an art 🖼️

Hi Non-Techies,

I like to think I’m a pretty good prompter (what a weird brag this would’ve been four years ago). I’ve even curated a massive library of prompts for AI Academy members.

But today’s headline tool is becoming a stalwart in my list of “nifty AI hacks”, so I thought I’d share it with you, too.

I’ll also update you on ChatGPT’s newly announced parental controls, and finish with a nod to my fave AI tool, which is now able to remember who I am. If you know, you know. You know?

Right, let’s crack on.

Keep it light, clear and sassy.

We’re about to get meta. Not in a Zuckerberg way, but in the traditional sense of the word.

OpenAI has released a little chunklet of software that makes your prompts better than they were before. Bafflingly, they haven’t given this a fun, sci-fi-inspired name like “The Optimonster” or “The Promptinator”, which feels like a missed opportunity.

They haven’t even named it after a fruit.

Still, we move. This thing is pretty good. You just type your prompt in the window:

Then hit the big “Optimize” button below. And hey presto, you’ve got a (theoretically) better prompt than you had before:

Role and Objective

• Create a light-hearted, jargon-free title for a newsletter section discussing OpenAI's new prompt optimiser, geared towards a non-technical audience with a sassy undertone.

Checklist

• Begin with a concise checklist (3-7 bullets) of what you will do; keep items conceptual, not implementation-level.

Instructions

• Generate section titles that are clear and accessible for non-tech readers.

• Inject a playful or slightly sassy tone—avoid technical language or jargon.

Context

• Newsletter audience: non-technical readers.

• Section topic: OpenAI's new prompt optimiser; the section debates its effectiveness.

• Newsletter writing style: light-hearted, occasionally sassy.

• Out of scope: Technical deep-dives, industry jargon, formal or somber approaches.

Verification

• After proposing titles, quickly verify in 1-2 lines that each is easily understood, jargon-free, and aligns with the intended tone. If any do not, self-correct before output.

Output Format

• Provide 3-5 potential section titles in a markdown list.

Verbosity

• Concise; only titles, no further explanation.

Stop Conditions

• Deliver section titles only when all criteria (audience, tone, and clarity) are met.

Persistence

• If unsure, default to maximum simplicity and humor; clarify edge cases in appended notes if applicable.

Set reasoning_effort=medium based on the need for creativity and careful tone alignment; keep reasoning internal unless explicitly requested.

Naturally, I fed ChatGPT-5 both of these prompts. Here are the answers:

Well, this didn’t go as I hoped. By copying the optimised prompt verbatim into ChatGPT, we get some good answers, but we also get some of the instructions too. Who in their right mind would name a newsletter section about a prompt optimiser “Keep it light, clear and sassy”? Certainly not me.

This tool will probably improve with time, and to be fair, other experiments my team and I have done have yielded better results. Give it a go by clicking the button below:

Oh, and if you like this sort of thing, we’re developing a prompt optimiser at AIFNT that I think is going to be way better. It’ll be free to use for AI Academy members:

ChatGPT’s new parental controls.

OpenAI is rolling out better parental controls over the next month. Do you know what, I’m just going to copy and paste the next bit from their blog post, because y’know, we’re all about efficiency in these parts.

Within the next month, parents will be able to:

• Link their account with their teen’s account (minimum age of 13) through a simple email invitation.

• Control how ChatGPT responds to their teen with age-appropriate model behavior rules, which are on by default.

• Manage which features to disable, including memory and chat history.

• Receive notifications when the system detects their teen is in a moment of acute distress. Expert input will guide this feature to support trust between parents and teens.

These are all steps in the right direction, and like a lot of OpenAI’s updates (and AI-related updates in general), they’ve already stressed that these steps are “only the beginning”.

As far as we’re concerned at AIFNT-HQ, more parental control over how young people use ChatGPT can only be a good thing.

Will Claude remember me now?

Anyone familiar with the lore of AIFNT will know that I’m a big fan of Anthropic’s Claude. For me, it’s the best AI tool for copywriting, hands, feet, knees and elbows down.

But I’ll be honest: The love wasn’t reciprocal. Using Claude was sort of like getting to meet my favourite podcaster. I feel like I know everything about them, and gush about them to everyone I know, but they still have no idea who I am.

Well, that’s about to change. Claude now has a memory, just like ChatGPT. It means it can recall past conversations and exchanges we’ve had, using that information to improve and personalise its responses.

It hasn’t been rolled out to free users (yet), so you’ll need to be a Pro user ($17/pm) to leverage this new feature. Click the button below to watch Anthropic’s 30-second announcement video:

I’ve made it onto the big stage!

Eek, I’m headlining at one of the UK’s biggest conferences, Technology for Marketing at the Excel in London on Thursday, 25th September. 

If you want to come and see me speak (and stop my terrible fear of having a headline stage and an empty audience coming to life), I’m on stage at 1:30-2:15pm.

My talk is about how I drove $75m of client sales using a $50/mo tool, and how I earned my favourite ever testimonial: “the most intelligent version of this type of campaign we've ever seen" - by Microsoft.

PLUS I’m doing a meet and greet in the Influencer Zone from 3:30pm. Bring me your AI issues and let’s do some delightfully cringy selfies together!

Best of all, tickets are free if you’re an in-house marketer:

Okay, that’s all for this week. If you have thoughts about today’s newsletter (good or bad), please share them with the poll below. It really helps us to improve.

See you next week,

Heather and the AIFNT team.