Hi Non-Techies,

I’m hosting a FREE, ~48-minute masterclass called “Become a ChatGPT Power User”.

It’s at 3pm GMT/10am EST on Thursday, 4th December. I’d love to see you there - just click the button below to register:

You’ll learn stuff like:

🤔 Which subscription is right for you?

✍️ How to write a REALLY good prompt (and why most prompts aren’t up to scratch).

📊 Deep Research and Advanced Data Analysis.

🧠 How to use ChatGPT’s other features, like Custom GPTs, Projects and Connectors (super handy!).

🤖 What the heck is an Agent?

It’s designed to take you from a casual ChatGPT user to a power user in under an hour.

A brief hiatus.

We’re taking a newsletter break until the end of the year, but we’ll be back with some Christmas presents, like an AI-powered Santa.

See you on December 4th? I hope so.

Heather and the AIFNT team.

Hi Non-Techies,

I’m back from a week of being disconnected from virtually every digital companion in my life (which, when you’re really into AI, is quite a lot of uninstallments).

And you know what? It was great. I love AI more than most people, but switching off from it (and everything else digital) every now and again is good for the soul.

Anyway, despite my strongly worded letter, AI didn’t stop just because I did. Selfish, really. Plenty has happened in my absence - let’s have a snoop.

Date for your diary 🗓️

Our final Become an AI Trainer 3-day Online Bootcamp of 2025 is happening on:

4:00 PM - 5:30 PM (GMT) on Wednesday, 26 November.
4:00 PM - 5:30 PM (GMT) on Thursday, 27 November.
4:00 PM - 5:30 PM (GMT) on Friday, 28 November.

In 3 live, action-packed sessions, we’ll cover:

🤖 The Booming AI Training Market.
🤖 Building Powerful Training Programs.
🤖 Advanced Techniques and Tools.

After 750+ attendees to date, we know Bootcamp gives you the tools to start your AI training business.

»Head here for more info and to save your space for £72/ $97«

Deep dive: Canva’s big AI update 🎨

Log in to Canva and you’ll see a big banner at the top that says “Watch the Canva Keynote”.

I’ve watched it so you don’t have to. Here are the big takeaways:

• AI assistance is everywhere in Canva, and you’re going to start seeing it everywhere. From editable AI-generated elements (like images, graphics, videos and 3D renders) to prompt-based design help.

• Video 2.0 is here. It has loads of fancy AI-powered editing and creating capabilities.

• They’ve released some other big features - most notably Email Design and Forms.

Putting it to the test

I’ve been playing with Canva’s new AI functionality since my return from the digital abyss. It hasn’t blown me away, but I can see the potential.

Here’s an example of it in action:

Having clicked the “Ask Canva” option on the design for this newsletter’s header image, this window pops up:

I can ask it anything, but there are some pre-written suggestions too. I selected the “Suggest improvements” option and received advice like this:

• Tighten line break. Keep “Canva’s new AI tools” on one line or split as “Canva’s new” / “AI tools” with balanced lengths to avoid an awkward pause.

• Increase contrast. Use a brighter text color like white or light cyan against the darker background, and add a subtle Shadow or Lift for legibility over the image.

• Unify imagery. Pick one background: the binary blue or the purple/gray gradient. Mixing both dilutes focus. Match text color to that palette for a cohesive look.

Okay - it hasn’t hit the mark here. Incidentally, “Ask Canva” won’t actually apply these tips for you. It’s just there for advice. You have to action the changes yourself.

Next, I asked it to rewrite the text. This is what it came up with:

Again, I don’t love these. They’re very generic and don’t actually mention Canva at all.

I’ve also been playing with AI-generated elements, which have proven useful when I can’t find a particular image or graphic. But like any image-generation tool, it needs re-prompting occasionally.

My verdict

I’m a big fan of Canva, and I’m pleased to see that it’s taking a proactive approach to incorporating AI into the platform.

But as a Non-Techie, I want Canva to be as simple as possible, and for me, that means being able to use natural language to create anything I want. It’ll get there eventually, but it’s still got a way to go.

Claude to be added to Excel 😍

Anthropic (Claude’s parents) will add Claude directly into Excel, which will answer a question I ask myself way too often: “Why is there an AI tool for everything, but I still can’t figure out how to read this spreadsheet?”.

It’s in Beta at the moment (you can join the waiting list here), but it’s important enough that I had to give you a heads up in advance. Here’s what Anthropic claim it’ll be able to do:

• Seeing #VALUE! is triggering. Claude will be able to diagnose and fix errors.

• Test hypothetical scenarios without breaking everything. “Claude, what if we doubled the conversion rate in cell D5?”

• Ask Claude for editable templates or to build something from scratch.

• And ask Claude basically anything at all about your spreadsheet.

From my experience, Non-Techies and spreadsheets don’t pair well together. Who better to fix that than Claude? We’ll test it once it’s live and let you know if it delivers on its promises.

Some more interesting stuff we found this week:

👀 Our 12-week Copilot course teaches you how to go from complete novice to the go-to Copilot pro in your office.

🎥 Google releases a TV ad made totally from AI.

💰 OpenAI is considering an IPO next year, with a suggested valuation of $1T.

Right, I'd better get back to it. I reckon if you printed out the number of unread emails I’ve come home to and sent them to my house, it’d be like that scene in Harry Potter with the envelopes coming through the fireplace.

See you next week,

Heather and the AIFNT team.

Hi Non-Techies,

Autumn? Meh. Winter? I can take it or leave it. But release season? Yup, that’s a season I can get into.

And in case it wasn’t obvious by now, we’re right in the middle of it. Every big AI tool is rolling out new releases, and for an AI enthusiast like yours truly, it’s like Christmas has come early (but y’know, without the annoying songs).

Today, I’m filling you in on the latest releases from some of the big players: Copilot and OpenAI. Let’s not dilly-dally.

Copilot wins the collaboration race 🏆

Until now, using a chatbot has been a lonely experience. It’s a single-player game - collaboration happens between you and your tool of choice.

But that’s about to change.

Copilot (Microsoft’s version of ChatGPT) now supports collaborative sessions, which means you can invite other people into your chatbot conversation for AI-assisted team brainstorming.

It’s called “Groups”, and it’s the first time that a big AI chatbot has given us the option to turn chatbot sessions into a team event.

Groups is one of a dozen AI updates released by Microsoft (I’ll cover four others I’m excited about in the next section), and it marks the latest step in a journey that has taken Copilot from a B-team AI tool to one of the best out there.

At AI for Non-Techies, we’ve just released a 12-week Copilot training course. It’ll turn you from a novice into the AI authority in your workplace.

And hey, if you don’t want to learn more about Copilot for the functionality, do it for Mico, their new amorphous blob:

Mico is the Copilot mascot. Definitely an upgrade on Clippy.

Four more need-to-know updates from Copilot 🫡

1) Connectors: Copilot now offers integration with external data and services (like your files in cloud storage) via “Connectors”. The demo included someone connecting Copilot with their Google Drive, which I didn’t expect, but it shows they want this feature to be genuinely helpful. There’s also a Connector for my beloved Claude!

2) Health-Related Queries: Here’s a pretty crazy stat: ~40% of users ask Copilot health-related questions every week. Because of that, Copilot’s health-answering capability is…well, healthier. They’ve teamed up with Harvard Health Publishing to provide better answers.

3) Study Mode: Much like NotebookLM's (probably superior) version, Copilot now has study-style tools (quizzes, flashcards) for interactive learning.

4) “Real Talk” interface: Ever feel like AI chatbots are a bit too…agreeable? Copilot now has a feature which mirrors your conversational (read: blunt) style and offers more candid responses.

Deep dive: ChatGPT’s new browser 👀

Meet Atlas: OpenAI’s new ChatGPT-powered internet browser.

The slightly more techie amongst you will know that Atlas isn’t the first fusion of an internet browser and a chatbot. Perplexity launched Comet back in July, and Google added some Gemini goodness to its Chrome browser recently, too.

But Atlas is exciting, because whenever OpenAI launches something big, it tends to set a new benchmark.

And in this case, Atlas does feel like the new benchmark for AI-powered browsers. It’s a sneak peek at the future of AI-assisted internet browsing.

Obviously, I gave it a try.

Atlas’s onboarding experience was pretty nice. I had the option to import my data from another browser (like bookmarks and stuff), and then I was asked whether I was happy to have memory switched on.

Memory is a cool feature - it means I can ask it questions like “You know those polkadot ear muffs I was browsing last week? Can you find them again for me?”.

Atlas’ homepage looks like ChatGPT, but with your tabs at the top.

I had to remember that the search bar isn’t Google, because I’m so used to just typing something like “AI for Non Techies” into my browser (probably with loads of typos) and getting a Google search result. That will take some getting used to.

But the biggest (and most useful, in my opinion) difference between Atlas and other internet browsers is the “Ask ChatGPT” button at the top-right of your window. You can ask it about anything you’re browsing and get instant answers.

So, if you’re on the AIFNT website and want to know if we’re legit, you can get ChatGPT to run a background check whilst you continue to browse:

Or maybe you’re browsing a holiday destination. You could ask ChatGPT about the weather in December without needing to specify where you’re referring.

What else can it do?

Atlas can also perform tasks for you. As the intro video demonstrates, you can ask it to shop for you, edit emails and browse on your behalf.

I’m pretty impressed. Like most of the stuff OpenAI puts out, Atlas is clean, no-nonsense, and intuitive in a way that won’t scare off Non-Techies.

But it also feels like a sign of things to come. It’s not hard to imagine that soon everything we do online will be AI-supported, from our Christmas shopping through to how we consume and process content.

Some more interesting stuff we found this week:

🚫 Bryan Cranston - as in, Breaking Bad Bryan - has released a joint statement with OpenAI after his voice and likeness were used on the Sora 2 app without his permission. OpenAI has committed to tightening the guardrails.

🎬 Speaking of Breaking Bad, Netflix is going “all in” on AI, according to their latest earnings report.

🌎 Google has revealed Google Earth AI, which combines Google Earth with fancy pants AI models to help tackle environmental risks, like flooding and wildfires.

Okay, time to design an amorphous blob for AI for Non Techies - apparently that’s the trendy thing to do during release season.

See you next week,

Heather and the AIFNT team.

Hi Non-Techies,

There have been loads more big AI announcements and releases over the last week - it feels like every day something massive is happening at the moment.

Having finally recovered from Monday morning’s tech blackout (what a way to start the working week, am I right?!), I’ve done my best to compile the most interesting and (hopefully) Non-Techie-relevant news for your perusal.

Keep reading for ChatGPT’s foray into erotica and an intro to our new Copilot course.

The great switchover 🔄

For years, Techies have been warning us about the great switchover. What started as the internet equivalent of a doomsdayer on a street corner quickly became a legitimate theory.

Now, that theory has become a reality - there is more AI-generated content on the internet than human-created content.

The study comes from Graphite, which studied 65,000 articles across a five-year stretch.

The astute amongst you will have noticed that the trend appears to be plateauing, though. Whilst there was a burst of AI-generated content soon after ChatGPT was released, the curve has steadied in the last year or so.

For me, this is a sign that whilst AI-generated content is easy to produce, it isn’t necessarily having the impact that its publishers had hoped. I mean, are you reading AI-produced content regularly? Probably not.

It’s another reminder for us Non-Techies that whilst AI adoption has been aggressive, it’s here to sit alongside human-created content, not replace it altogether.

ChatGPT is about to reach adulthood 😳

Last week, Sam Altman revealed that OpenAI will soon be able to “safely relax the restrictions” that have curtailed ChatGPT’s adult-only content.

ChatGPT will also get a personality revamp after complaints that ChatGPT 5 is a bit too…wooden.

As Sam explains in his tweet, this is part of the “treat adult users like adults” policy at OpenAI. So if you want erotica from ChatGPT, you can have it from December.

Hey, I’m not here to judge how people want to use their AI tools. For me, the big talking point for Non-Techies here is how workplaces might interpret these new developments.

Sam Altman was careful to stress that users “won’t get it unless they ask for it” - it’s not like you’ll ask for a marketing plan and it’ll reply with something that wouldn’t be out of place in 50 Shades of Grey.

Still, I feel like this could be used by workplaces to enforce restrictive rules that limit our capacity to use AI to its full potential. Time will tell. What are your thoughts on this?

Claude gets some new Skills 📈

Anthropic has given Claude some new skills. Ingeniously, they’ve called them Skills.

What are Skills?

Skills are customisable instructions that Claude can use again and again. As their article reads, “Think of Skills as custom onboarding materials that let you package expertise, making Claude a specialist on what matters most to you.”

You upload them as a folder into Claude’s brain, and then it can call upon that Skill when delivering work.

What’s a use case for Skills?

I used a marketing plan as an example in the last section, so I’ll use it again here. Let’s say you want to create a marketing plan for your business. You could create a Skill in Claude that shares your brand guidelines. A different Skill could share your revenue goals for Q4. A third Skill could share the marketing assets you have at your disposal.

Claude can combine and “stack” your Skills to create something totally cohesive.

Crucially, Skills are repeatable. They use the same format everywhere. So if you’ve uploaded your brand guidelines once, you can call upon that Skill again and again for decks, ad creative, reports and just about anything else you can think of.

Skills are available to Pro, Max, Team and Enterprise users. This 70-second video demonstrates how they’re used nicely 👇

Copilot’s huge improvements 🚀

Long-time readers of this newsletter will know that I haven’t exactly been a cheerleader for Microsoft’s flagship AI tool, Copilot, in recent years.

It used to be way behind rival tools like ChatGPT, Gemini and Claude.

But that’s not the case anymore. Moaning about Copilot is so 2024. It’s come on in leaps and bounds since then.

I did a webinar about these improvements earlier this week, where I talked about five features I love about Copilot in 2025:

Chat - proper prompting and nifty prompt tweaking tips
Notebook - an interactive knowledge bank
Agents - build your own chatbot (no tech skills needed)
Create - design quality images in your brand style
Apps - a handy helper in Outlook, Excel, Word, etc.

It coincides with a new course we’re launching at AIFNT HQ, the 12-week Ultimate Copilot for Work course. Go from nervous dabbler to saving 10+ hours every week and enrol today 👇

Some more interesting stuff we found this week:

🎥 Google launches Veo 3.1 - the latest instalment in its video production and editing tool.

🤝 Anthropic integrates Claude and Microsoft 365 for Enterprise users.

Right then, I’m off to pack Claude with as many Skills as I possibly can.

See you next week? I hope so.

Heather and the AIFNT team.

Hi Non-Techies,

I feel like AI has been in the news a lot this week, but it’s hard to tell because I’m surrounded by it all the time. Perhaps there are gardeners out there who feel as though the whole world is talking about whether their Chrysanthemums should’ve flowered by now.

But then I’ll stumble upon an incredible graph like this:

And I’ll think, “Nope, this isn’t just me. The world really is talking about AI all the time.”

Nowhere is that clearer than in my AI Academy. If you want to join a community of AI-curious Non-Techies from all kinds of professional and personal backgrounds, click the button below to start your 7-day free trial:

The AI app to rival TikTok 📈

Some of you might know that when the ChatGPT app was released, it reached 1,000,000 downloads faster than any other app in history.

Well, that record has just been broken. Thankfully for OpenAI, it’s been broken by another OpenAI product, Sora.

Sora 2, launched inside the Sora app this month, is a bit like TikTok, but every video on there is AI-generated. Users are encouraged to create their own 10-second AI videos using prompts, and can feature themselves, their friends (with permission) or some publicly available figures (like OpenAI’s CEO, Sam Altman - here’s a Sora video of him eating a cloud).

Who hasn’t thought about Sam Altman chomping on a cloud?

As you might imagine, the response has been mixed. Like so many AI releases, Sora 2 is startlingly impressive, scary and, at times, problematic, all at the same time.

Issues range from copyright infringement (the app is filled with intellectual property from films, TV shows and games) to Robin Williams’ daughter having to ask publicly not to have AI-generated videos of the late actor sent to her.

For me, there’s a lot of creative, funny and occasionally inspiring content coming out of Sora 2, but I’m also mindful of how easily it could be weaponised for online abuse. OpenAI will have to adapt quickly to curtail any misuse of such a powerful and popular tool.

It’s only available in the U.S. at the moment, with no clear timeline for a UK/EU release.

Are you rude to your AI tools? 😳

I tend to use “please” and “thank you” when I’m talking to my AI tools, especially if the tool is Claude. I find that a lot of AI consultants (certainly the ones that we work with at AIFNT, anyway) are in the same boat.

I’d probably need a separate newsletter to explain why, but if I had to guess, I’d say it’s 49% manners, 50% habit and 1% fear of a Terminator-style reckoning at some point in the future.

But a recent study has shown that rude prompts produce better results than polite ones. From the report:

“Contrary to expectations, impolite prompts consistently outperformed polite ones, with accuracy ranging from 80.8% for Very Polite prompts to 84.8% for Very Rude prompts.”

Here’s an example of a rude, “Man Men-style” copywriting prompt (possibly not safe for work, depending on your work).

How to tell an AI image from a real one 👀

A few newsletters ago, I introduced you to Nano Banana, Google’s scarily powerful AI image generator.

It’s so good, in fact, that the question of image authenticity - i.e. what’s real and what’s AI - has never felt more important.

Some smart Redditors may have discovered an answer. They’ve found that if you oversaturate an image generated in Nano Banana, it reveals a watermark. Here it is:

It reminds me of those optical illusions where, if you blur your eyes, you can see the outline of an elephant or something (I tried and, sadly, no elephant was found).

This is a really important topic. I want to live in a world where AI-generated content is easy to identify, not just for my own sanity, but because there are all kinds of dangerous implications if the line between real and AI content gets too blurred.

Whether or not these watermarks are the solution is another question entirely (presumably, it’s possible to remove them?), but for now, it’s comforting to know they exist.

A game-changer for marketers 🤝

I stumbled upon this clever use of LLMs (Large Language Models - like ChatGPT) the other week and thought, “I know someone who will find this interesting”.

The idea is to create a “synthetic consumer” to better predict (with 90% accuracy!) whether or not a customer will buy a product.

It’s from a paper titled “LLMs Reproduce Human Purchase Intent via Semantic Similarity Elicitation of Likert Ratings”. Pardon? In Non-Techie speak, here’s how it works:

1) We give ChatGPT a demographic profile of a customer.

2) We give this “customer” a product and ask it for its thoughts on this product.

3) We then use another AI to rate the thoughts of this “customer” based on their probability of buying.

This is pretty cool. Imagine we want to launch a new product or service for our customers. Using AI, we can effectively create a testing group to see whether it’s viable, without actually having to launch the thing. Companies spend BILLIONS on this stuff.

If you’re feeling extra Techie, you can read the full paper (Warning: it’s very academic):

Some more interesting stuff we found this week:

🧠 This 10-minute tutorial video on how to combine NotebookLM with Perplexity for extra usefulness is really good.

🗑️ You can now permanently delete your ChatGPT chat history. ChatGPT conversations were being preserved as evidence in a lawsuit with the New York Times, but a judge has ruled they can be permanently deleted once more.

Okay, I’m going to spend the next week trying to muster up some rudeness to use in my prompts.

Starting from now - bye. Go away. Whatever.

(I just can’t - thanks for reading and please come back next week xx)

Heather and the AIFNT team.

Hi Non-Techies,

I’ve spoken about mental health in this newsletter a fair bit before. Some of our long-time readers might remember this email, in which I talked about how my custom anxiety GPT once calmed me down during an anxiety attack in the back of a taxi.

Well, things move fast in AI (have I mentioned that before?), so today I’m sharing some more tools, resources, prompts and thoughts on how AI-powered mental health support is evolving.

With the exception of some of the tools (which have paid tiers), everything mentioned in this newsletter is free.

The AI for Mental Health webinar 🎬

I hosted a webinar last week about AI for mental health. On the panel were three experts in this department:

Jake Mills, the CEO of mental health charity Chasing the Stigma and their mental health support directory, Hub of Hope.

Nick Jemetta, an AI coach and mental health speaker.

Chris Reinberg, the CEO at Mindsera (and AI-powered journal).

The four of us spoke about our own mental health stories, but also discussed the tools we use (and avoid), some prompts and, perhaps most importantly, the risks and limitations to using AI for mental health.

If you’re the type that prefers to watch rather than read, you can watch the webinar here:

Three AI tools for mental health ⚒️

AI isn’t just for creating fun pics and drafting delicate emails. The development of AI-powered mental health support tools has been a fantastic application of AI’s immense potential.

Here are three tools we’d recommend for slightly different needs:

1) Headspace (feat. Ebb)

Headspace has been around since 2010, long before AI became the thing even your Nan talks about.

Predominantly, it functions as a meditation and sleep assistant, but its new AI co-pilot, Ebb, can do much more.

As Nick discussed in the webinar, he was impressed with the signposting to external, real-world services that Ebb provided.

2) Mindsera

I love Mindsera, in part because I think journaling is seriously underrated. It’s like going to the gym, but for your mind (way less sweaty, most of the time).

But Mindsera isn’t just a journal. As its homepage declares, it’s a journal that reflects back. It encourages you to explore particular feelings, invites you to dig deeper, and guides you towards getting more out of your journaling.

Super simple on the surface, but with some features that make great use of the science of mental health.

3) Pi

Don’t worry, I’m not talking about the mathematical concept here. Pi.ai is a chatbot not dissimilar to Claude or ChatGPT, but with an emphasis on having empathetic, supportive conversations.

If you’re wrestling with something and you need a conversation about it, Pi is a great option.

I know a lot of people using Pi, and its simple, elegant interface makes it super Non-Techie-friendly too.

The risks of using AI for mental health 🧐

“Any AI tools you’re using outside of your core, you wanna be looking for whether they’re talking about safety”.

I couldn’t agree more with this point from Nick during the webinar. As with all burgeoning tech, the speed of Techie progress in AI is outpacing the speed with which effective safety measures can be put in place. That’s why I love it when tools are built with a safety-first approach from the start.

I’m using “safety” as a bit of a catch-all term here. In the literal sense, it means ensuring that the tools and resources you’re using are dedicated to safeguarding your well-being above all else, signposting to other services when needed.

But it can also mean the way your data is used, the amount a tool hallucinates or the baked-in biases that are so common in even the most sophisticated AI software.

And not every tool is getting this right. I tested Replika, a tool designed for young people that lets users create avatars with which to discuss mental health issues. It’s a great idea in theory, but in practice, the responses to my test questions were concerningly dismissive in my view:

As Jake from Chasing the Stigma explained, AI awareness is increasing in the UK, which is amazing, but services are stagnating. AI tools can bridge that gap, but that’s a huge responsibility for burgeoning tech, and it’s critical they get it right.

Remember: Every responsible AI tool for mental health will have a dedicated, comprehensive section on its website about safety.

Some more helpful resources:

💝 My mental health prompts sheet. These work for me, but feel free to customise them for your own needs!

🧠 My custom anxiety GPT. Again, this was designed for me, so…

🔧My custom GPT builder tool, for those of you who want to build your own custom GPTs.

I love that AI is being used to provide easy-to-access, affordable mental health support to anyone who needs it, and I’m excited to track its progress.

But whilst AI tools can provide a great starting point, it’s critical that they’re built responsibly, with signposting to other services as standard. Anything less, and the consequences could be devastating.

Thanks for reading - my replies are always open!

See you next week,

Heather and the AIFNT team.

Hi Non-Techies,

I headlined Technology for Marketing at the Excel centre last week, which I’m mentioning mostly because it gives me an excuse to share this awesome pic with you:

I’m the one stood up at the front.

Becoming an AI trainer is the best thing I’ve ever done, so I built a 12-week bootcamp that’ll turn you into an AI trainer too. Click the button below to learn more:

In today’s newsletter, I’m sharing a useful prompt, a tool that can actually clone my brummie accent, and discussing the US Government’s AI tool of choice.

Bibs on, let’s dig in.

Finally, an AI-generated brummie accent 🎤 

I tried Synthesia’s new voice clone tool the other day and I was really impressed.

AI voice cloning isn’t new, but I’ve tried plenty of tools, and none of them could ever do my voice. If you’ve never heard me speak, I’m from the rough end of Birmingham (the UK version), so you know, I don’t exactly sound like the Queen.

Synthesia’s voice cloning tool is the first I’ve discovered that can do it. It’s not perfect, but it’s pretty darn good.

I was also really impressed at how easy Synthesia made it for me. A lot of voice cloning tools learn your voice by asking you to speak solidly for two minutes, but have you ever tried to do that? Either you start speaking gibberish, or you read, and none of those are particularly great examples of how you actually talk.

Synthesia walks you through the process, giving you prompts to ensure you’re not lost for words.

Oh - and their trust centre is epic.

Why does this matter?

If I can clone my voice, I can start to scale my content output. I could create podcasts (and in loads of different languages), training videos for my team, even YouTube videos: It all becomes easier with great voice cloning.

It’s only available on the Enterprise plan at the moment, but it’s a really promising development in this space.

A prompt you might find useful 🧠

Chatbots have come a long way, but even my beloved Claude can still feel a bit…well, a bit like AI is doing the writing. If feedback sessions in 2025 had a bingo card, “It sounds like AI wrote it” would surely be on it somewhere.

As it turns out, that might be a prompting issue, not a tech issue.

We tested this prompt in the AI Academy yesterday, which claims to be able to coax any chatbot into writing like a real human.

The results were pretty good. It still needed a bit of further prompting to get it totally right, but some of that comes down to personal preference.

Which reminds me: If you ever need to refine a prompt with extra prompts to get what you want, be sure to ask ChatGPT (or your chosen AI chatbot) to give you a new prompt at the end that factors in all of the new information you’ve given it. It’ll save you the trouble next time!

GrokAI for $0.42 👀

Sorry, but not for you. Elon Musk is only extending this offer to those who need it most: The US Government.

It’s the latest in an intriguing battle amongst AI companies - namely, OpenAI, Anthropic and Grok - to become the AI provider of choice for the US Government.

OpenAI and Anthropic have already offered full, enterprise use of their platforms for a dollar. Not to be outdone, Musk has gone in at $0.42 for a year and a half of usage. I can only assume the $0.42 is an in-joke, but I won’t discuss it further because of spam filters.

On first glance, it’s kinda hard to see why these companies are so keen to charge so little. Clearly, each sees a big advantage to being the government’s go-to AI tool, presumably because it gives them even more legitimacy. You can read the full story here if you’re intrigued by this bizarre bidding war.

Some other cool stuff we found this week 😎

📲 OpenAI have released ChatGPT Pulse: an assistant that can send daily updates, reminders or tips based on your calendar, other integrations and manual requests. It’s only available at the Pro level, mind you.

💌 Perplexity has launched a “personal assistant for your inbox” that organises your emails, drafts replies and organises meetings. Like ChatGPT Pulse, it’s only available at the top subscription tier ($200/pm!). Pretty steep.

🌍 Al Gore (that’s Al as in Albert, not Artificial Intelligence) has released an AI tool that can monitor global pollution, using 660 million data sources. Read more about it here.

Okay, I’m off to ask Anthropic if they’ll give me Claude for $1. Hey, if you don’t ask, you don’t get.

See you next week,

Heather and the AIFNT team.

Hi Non-Techies,

I’ve banged on about NotebookLM before, but last week I found a great resource for those of you who aren’t yet using it to its full capacity. Today I’m sharing it.

I’ll also be talking about some of Zoom’s new features, reflecting on last week’s Zoomtopia, introducing you to Notion’s new, super smart AI agent and sharing some other cool stuff my team and I have stumbled upon over the last week.

If this newsletter were a cheeseboard, it’d be a full spread. Baked camembert in the middle, breadsticks, carrot batons, the works.

We demo a lot of these emerging tools in live sessions over at the AI Academy. It’s the best way to get a proper feel for how they actually work (and whether they’re as good as they claim). Want to join one? Click the button below for a 7-day free trial.

Say goodbye to bedhead on Zoom 😂

It’s been a big week for Zoom, so this is the first of two sections about their AI offerings.

Live AI avatars are coming to Zoom. It’ll let you create a photorealistic version of yourself that will sync up to your real-time movements, but will spare you any blushes if you’re not quite “camera-ready”.

Zoom users (Zoomers?) can already use AI avatars for pre-recorded messages, but this is the first time you’ll be able to use them in real-time video calls.

The clips function in Zoom allows users to generate an AI avatar for pre-recorded messages.

You can also change your avatar’s outfit, so if you want to look more professional but can’t be bothered to change out of your pyjamas, your AI avatar can don the power suit on your behalf.

The natural concern here is fake avatars, but Zoom has a solution for that already. To prevent people from pretending to be someone they’re not, they’ll use real-time authentication to ensure the person on cam is the same person as their AI avatar.

The best NotebookLM resource I’ve found 📍

You know when you’re shopping for a car, and then you start seeing it everywhere?

Google’s NotebookLM is a bit like that. Since I’ve started using it all the time, I keep seeing it in the wild. And it’s always stuff like this:

This was recommended to me on Reddit the other day. I don’t even subscribe to the subreddit.

Anyway, the most helpful resource I’ve found to get you acquainted with NotebookLM is, shockingly, their own introductory guide. Not only is it a great explainer, but it’s also a living, breathing example of what NotebookLM can do.

If you’re not familiar with it yet (or its new features, like quizzes, flashcards, upgraded reports and additional audio overview formats), click the button below to find out why so many of its users shout about it in capital letters on the internet:

“It basically does your job for you” 🤖

People who use Notion tend to love Notion. It creates megafans, not unlike NotebookLM…or The Beatles. If you’re unfamiliar, it’s a workspace, primarily for creating documents and databases.

Last week, they announced Notion 3.0, which, in layman’s terms, means “Notion but with agents”.

It’s the latest in a trend of tech platforms releasing super-powerful agents (we discussed Replit’s agent in last week’s newsletter) that can essentially do everything that a human can do on the platform, but way faster.

As the press release states: “Anything you can do in Notion, your Agent can do too. The busywork that fills your day can now be done in minutes.”

I nabbed this off their website.

It feels like the AI tools that can do, rather than just tell, are going to be the main focus in AI over the next few years.

Which, let’s not forget, should be music to our Non-Techie ears. The more that AI can do for us with a simple prompt, the less we have to worry about the technicalities.

If you’re feeling curious, you can read Notion’s press release here:

Some other cool stuff we found this week 😎

🤷 Can we ever trust online video again? Higgsfield, who are always at the forefront of AI visuals, have released Lipsync Studio, which can make anyone look like they’re saying anything. It’s free to use at the moment, and this tweet (it’s full of profanity, so don’t blast it out in the office) shows it in action. Kinda scary, very amazing.

⏰ This copy+paste addition to any prompt ensures your chatbot knows what date and time it is. Great for time-sensitive research, for example.

🌍 Google have added Gemini to their Chrome browser, with loads of new features (like asking Gemini to clarify complex information on a webpage). Soon, they’ll be introducing an agent that can “book a haircut or order your weekly groceries”. We’ll definitely do a proper section about this once it’s rolled out.

Zoomtopia: Event review [ad]

I attended Zoomtopia last week - did you go too? One of my favourite sessions was the Productivity track.

It was a really interesting panel at Zoomtopia with two completely different business owners - Mitch who runs a consulting business and Bernd who's head of IT at a huge engineering firm.

Here are my 3 key takeaways:

1. Content Repurposing
   Mitch records every client call and podcast interview, then asks Zoom AI Companion to turn those transcripts into LinkedIn posts, newsletter articles, even custom proposals. One conversation = weeks of content, without the need for an extra AI note-taker. Definitely one I’ll be trying.
   

2. AI That Actually Holds People Accountable
   Bernd showed how Zoom AI Companion can automatically catch when someone says "I'll handle that" in a meeting and turns it into an actual task…with reminders. This could save so much mental energy tracking what everyone promised to do.
   

3. One Platform Instead of Tool Chaos
   Both of them basically live in Zoom now - meetings, documents, tasks, even phone calls all in one place with AI helping everywhere. Bernd mentioned they even integrated their email and calendar. I think we’re all worried about having too many expensive AI tools - so the ones that offer a lot of use cases in one place will win.

You can register to watch the whole Zoomtopia event on demand right here:

Okay, excuse me while I pick out some outfits for my Zoom avatar. It’s like having a Tamagotchi, but with your own face. And with that rather disturbing image burned into our brains, I’ll see you next week.

Heather and the AIFNT team.

Hi Non-Techies,

You’re getting two newsletters this week, because I missed one last week. It’s like Christmas has come early, if Christmas means an abundance of AI news geared towards Non-Techies.

I’ll be talking about Claude today. What’s new? I’ll also be discussing Replit’s fancy new agent, inviting you to a Zoom call and sharing some political news from Albania. Obviously.

If a weekly email just isn’t enough AIFNT for you, you’re probably the sort that would enjoy our AI Academy. It’s £25/pm, but you can try it for free if you click the button below:

“Claude, turn this section into a PowerPoint presentation”

You know when Season 2 takes so long to come out that you have to watch a recap of Season 1 just to remember what happened? Well, that’s what I’m about to give you.

So, a quick recap if you’re new around here: Claude is an AI Chatbot made by Anthropic. It’s the best tool for writing (in my opinion, at least), and it’s super Non-Techie friendly, a bit like ChatGPT.

But unlike ChatGPT, it lacked some of the fancier functionality that would’ve made it a real game-changer. Like, up until about a month ago, it didn’t have a memory, which meant it couldn’t reference previous chats.

It’s having a good summer (is it still summer?) though, because it’s just had another big update: now it can create and edit Excel, Word, PowerPoint and PDF documents.

So if you want to, I dunno, shove a load of messy data in there and tell it to create a spreadsheet, it can do that. Or maybe you want to turn meeting notes into a PowerPoint? It’ll do that too.

Pretty cool.

You can watch the announcement video here if you’re that way inclined.

The mega agent.

Sorry, but this section isn’t actually about a mega agent. As far as I know, the mega agent doesn’t exist yet.

But if there’s one thing that really sucks about the AI space, it’s the lack of excellent names. You create an agent that can create other agents, and what do you call it? Agent 3. This is a waste of near-Promptinator proportions.

First, some context. Replit is a tool that helps you create apps and websites using nothing more than AI and your ideas. Don’t worry, you’re safe: it’s designed for us Non-Techies, so much so that they actually have the words “No coding or technical skills required” underlined on their homepage. Phew.

They’ve just released a new agent called Agent 3, which tests and fixes your code. It’s apparently 10x more autonomous than their last agent (presumably called Agent 2?).

Bla bla, whateather Heather - give me the meaty bit.

I’m mentioning it here because Agent 3 might be the future of AI agents. It can:

• Understand your goals with natural language prompts.

• Work for 200 minutes on its own.

• Build other agents and automations.

That third one is the one that has my eyebrows scaling the steep incline of my forehead. An agent that can build other agents and automations is really powerful.

In real terms, let’s say I want to build a little app that gives me an update on everything happening in the AI industry on Monday mornings. Replit’s Agent 3 (i.e. the mega agent) can do that, building the automations needed to make it happen.

If this is a sign of things to come where agentic AI is concerned, we’re in for a real treat.

When your MP needs an update…

The Algerian government has appointed the world’s first AI minister. Not someone who looks after the government’s AI policy, but a minister created by AI.

Her name is Diella. She looks after the government’s procurement, and honestly, the more I read about it, the more I quite like the sound of her.

Here’s Diella, looking ready to procure.

Regular readers will know that my enthusiasm for AI is, in part, fuelled by an unshakable belief that it will champion and accelerate the power of the human brain.

But the logic behind creating Diella makes total sense. As Algeria’s Prime Minister, Edi Rama, explained, Diella will ensure government procurement is “100% incorruptible, where every public fund that goes through the tender procedure is 100% legible.”

Obviously, she’ll need some human help. But don’t we all sometimes?

Zoom’s New Agentic AI Features

I’m off to Zoomtopia tomorrow (Sep 18th for EMEA folks) to find out about a few of good old Zoom’s interesting new AI tools:

• Automatically scheduling meetings you mention in calls

• Drafting follow-up emails based on conversation context

• Identifying tasks from meetings without you having to ask

They're also demoing their new AI contact centre features - voice and chat agents that sound genuinely useful.

The event is completely free and virtual, specifically designed for small businesses.

Come join me:

[Ad]

Right, I’m off. If I don’t see you on Zoom tomorrow, I’ll see you next week right here.

Heather and the AIFNT team.

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:

I warned you about it getting meta.

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:

Results from the first prompt.

There seems to have been a misunderstanding.

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.

Hi Non-Techies,

Sorry I missed last week, I was away on holiday.

Me on holiday, presumably about to attempt a cartwheel?

I’ll tell you who wasn’t on holiday last week: Google. No beach-based cartwheels for them.

Instead, they were rolling out some big old updates to their image, language and NotebookLM tools.

In today’s newsletter, I’m going to take you back to primary school and do a bit of show and tell. But rather than a fossil or a novelty puppet (don’t ask), I’ll be showing you what Google’s new tools can do, and telling you how to start using them.

Btw, if you want to see live demonstrations of new AI tech in action, we host them every Monday in the AI Academy.

It’s only £25/pm to join, but you can try-before-you-buy with a 7-day free trial first:

“Put me in a room full of dogs”.

Apple. Blackberry. Raspberry Pi. If your exciting new tech isn’t named after a fruit, is it really exciting at all?

Well, Google’s new image editor and generator, Nano Banana*, could be called “Logistical Solutions” and I’d still find it exciting.

Firstly, it’s really good at maintaining “character likeness”, which is another way of saying “you can edit a picture of a person, and the person will still look the same”.

That goes for simple stuff, like this:

This took 8 seconds, by the way.

And more complex stuff, like this:

This one took 10 seconds to generate.

This is pretty impressive. Not only do I look like the same person in the AI-generated images, but details like my hair colour and the tattoo on my left arm are also really accurate.

I’m also blown away by how quickly it produced them.

*Incidentally, it looks like Google already regret calling it Nano Banana, because they’ve scrubbed it from the release statements. You’ll also see it referred to as Gemini 2.5 Flash Image. Not quite as catchy, but like I say, I’m excited regardless.

What happens when you blend images?

Remember that YouTube series called “Will it Blend?” that put random objects into a blender to see what would happen? I’ve been playing an image version of that with Nano Banana.

As you can see, I’m putting it through rigorous and important tests.

Google’s own blog announcement includes a better example:

Pick some images and watch Nano Banana blend them together.

And here’s one I found on X by Travis Davids, which combines 13 different images into one:

The best thing is that the headphones went on the dog.

Image blending isn’t a new thing in the AI space, but complex, fast image blending is hard to come by, and Google seems to have cracked it here.

You can access these image features for free in the Gemini App. They’ve even written an article that offers tips on how to get the best out of them.

Hola. Dos…cerveza? Por favor.

There’s nothing like going abroad to remind yourself of how monolingual you are. Google’s AI-powered live translation is here for people like me, stumbling my way through a beer order in Mexico, or people who need to have full-blown, complex conversations in over 70 different languages.

This video demonstrates it in action.

Note: This tool is nothing new. Google can even do live translation in Google Meet, so this is definitely an update, not a launch.

Google have also built a language learning tool that offers customised listening and speaking practice sessions based on your abilities and targets.

Whilst the latest demo looks like a faster, more fluid and more accurate version of Google’s real-time translation tool, the reception to this update has been less enthusiastic than their image tool updates. According to early users, it’s slower than the demo video suggests, and still prone to regular errors.

What if language was no longer a barrier?

This is the rather bold (and grammatically incorrect, according to Grammarly) question posed in the demo video for Google’s latest update to NotebookLM.

Last month, we mentioned the introduction of Video Overviews, which can create video presentations from simple inputs (like a PDF), and now it’s available in over 80 languages.

So if you need to create a video presentation for your colleagues in Japan, now you can.

NotebookLM is a bit of a fan favourite amongst its users, but, for reasons I can only attribute to branding, it still flies under the radar for most entry-level AI users. If you’re ready to join the AI hipsters, we’re big NotebookLM users at AIFNT, and we wholeheartedly recommend it.

After the letdown of ChatGPT 5, there was pressure on Google to deliver with these latest rollouts. And deliver they have. For Non-Techies like us, it feels like they’re the ones to watch now.

Right, I’m leaving you to continue my serious AI research. I’m thinking of blending a picture of a fun-loving parakeet with a picture of a fancy whisky glass. Will it be standing in the whisky glass or holding it? These are the questions I need to answer.

Incidentally, is there a Nobel Prize for AI breakthroughs?

See you next week,

Heather and the AIFNT team.

Hi Non-Techies,

Well, this is awkward.

ChatGPT-5 finally dropped 11 days ago, and instead of a celebration, we got a Reddit thread titled "GPT-5 is horrible" with 4,600 upvotes.

We were told we were getting an epic upgrade.
What we seemed to get was an AI having a proper personality crisis.

Let me explain what's really happening – and how to fix it.

The Great ChatGPT Personality Lobotomy

The biggest complaint isn't about intelligence (GPT-5 is genuinely, measurably smarter). It's about how it feels.

Users are saying GPT-5 talks like a "LinkedIn-approved ghost" and they're "genuinely grieving" GPT-4's warmth. One person nailed it: "Sure, GPT-5 might be a straight-A student, but GPT-4 was a friend."

Yup. OpenAI optimised for accuracy and accidentally performed a personality lobotomy. Where GPT-4 felt conversational, GPT-5 reads a bit like a corporate memo. It’s weird, and feels like a huge step back.

When a "PhD-Level Expert" Can't Count Letters

Then there are the technical blunders.

GPT-5 insisted there are three "b"s in "blueberry" and got confused about US states. This echoes past, well-documented problems (it couldn’t count the “r”s in “strawberry” a long time ago) - not exactly the expert-level performance promised.

Solid attempt, try again.

The culprit is a new routing system that decides whether to use "fast" mode or "thinking" mode. When it glitches, simple questions get routed to the wrong model, creating these sorts of facepalm moments.

The Subscription Swindle

They also got loyal paying users feeling properly wound up.

ChatGPT Plus subscribers suddenly found themselves capped at 200 messages per week for the good stuff, while their trusted older models vanished overnight.

As one Reddit user put it: "It feels like a downgrade branded as the new hotness."

They’ve got to be careful. They could alienate a massive base of paying users if they continue to offer very limited usage. The free version needs to be substantially worse, or why bother paying!?

Here’s How to Actually Make GPT-5 Useful

Before you give up entirely, some early adopters have sussed out how to make it work properly:

1. Fix the personality crisis

OpenAI added four personality modes because people hated the corporate tone. Go to Settings > Personalisation and pick Cynic (sarcastic), Robot (efficient), Listener (warm), or Nerd (enthusiastic). Or add your own in.

Here’s mine, if you’re interested:

2. Prompt like you're briefing a PhD student

GPT-5 is hyper-sensitive to instruction style. Instead of "write me a marketing plan," try "Break this into components, then create a structured plan for each." This is called ‘chain of thought’ prompting - literally share how you’d do it manually as a human. More specific = better results.

3. Master the three modes

You can now choose: Fast (simple stuff), Thinking (complex problems), or Auto (let it decide). This prevents those "blueberry has three b's" disasters.

4. Feed it bigger projects

GPT-5's context window is massive – upload entire documents instead of breaking everything into tiny chunks. It's designed for substantial inputs and genuinely can handle all the extra stuff.

5. Treat it like a conversation

Power users found it works best with back-and-forth. Start broad, then refine: "expand the opportunities section" or "make this more specific." It’s not just a ‘click and go’ thing.

There's Hope (I Think…)

The encouraging bit is that OpenAI is listening.

Within days, Sam Altman brought back GPT-4o (see pic below) and promised to make GPT-5 "warmer." They've increased rate limits and are working on better personalisation.

Ah - hello, old friend.

Plus, the fundamentals are solid – 45% fewer factual errors than GPT-4o, better reasoning and significantly improved accuracy. It's just the user experience that needed work.

I think a few more weeks of this active listening, and we’ll have ourselves a much stronger product.

Your CRM just learned how to talk 🤯

Tired of fighting your CRM for basic information? We've all been there - needing simple answers like "How many deals are closing this month?" but having to do 25 clicks, build custom reports, and practically perform a rain dance just to get basic data. It's unnecessarily faffy and frustrating.

Well, HubSpot and ChatGPT just solved this by partnering up so you can ask your CRM questions in plain English and get actual answers. 

To access, go to your ChatGPT: 

Click on your name >> Settings >> Connectors >> Hubspot >> Connect

You'll need paid ChatGPT and HubSpot accounts, but if you're done with the endless clicking, this is your solution. 

There's a live demo on Thursday, 21st August if you want to see it in action:

Right, it’s my birthday today, so I’m now off to go and eat yet more of that lovely Costco birthday cake (what do they put in it to make it so sickly but so addictive?!)

See you next week,

Heather and the AIFNT team.

Hi Non-Techies,

ChatGPT 5 has just been launched, and it’s all the AI world is talking about at the moment. I’m going to be a proper hipster* and talk about it next week instead, but I shared my initial thoughts on LinkedIn if you’re particularly eager to hear them.

It’s the summer holidays and, apparently, kiddies need entertainment now more than ever. Today’s newsletter is going to explore two ways that AI can help.

Spoiler: I’ll be sharing a free Google Doc of prompts at the end as well.

Before I get going, here’s a final reminder that my 3-day Become an AI Trainer bootcamp starts next week.

Join me digitally on 19-21 August 2025, 16:00-17:30 BST / 11:00 - 12:30 ET for three days of learning. Best of all, it’ll only cost you (or your company) £72/$97.

Just click the button below to register:

*You can also listen to this newsletter on vinyl, if you’d prefer.

Create a custom storybook in seconds 📚

Google’s Gemini launched a new tool last week called Storybook. It lets you create a personalised, illustrated story using a simple prompt. Naturally, I put it to the test.

If this isn’t the next Harry Potter, I don’t know what is.

Here’s my favourite page from the book, where our hero, Heather, overhears a man called Arthur moaning about AI in the park (an actual park in Birmingham, btw). Something that happens to me all the time, obviously.

If only I were so gentle and soft in real life.

Okay, so this is pretty cool. Here’s what I liked about it:

• It was super quick. Took less than a minute to generate.

• It’s free.

• Whilst it wasn’t quite a Stephen King novel, the quality of the story was really good, as were the illustrations.

There’s plenty you can do to customise this, from changing the reading level to using a photo to accompany the prompt. Not a bad way to spend an afternoon!

Choose your own adventure 🧭

Not to brag too much, but this is one of mine.

I used to love those choose-your-own-adventure books when I was a kid.

You know the ones:

“You’ve approached a bootcamp plug in the AIFNT newsletter: Do you turn back and run, or click and find out more?” - An extract from the most boring choose-your-own-adventure ever.

Whisper it, but apparently AI isn’t just for speeding up workflows, so I spent some time putting together some prompts so that you, the kids, the grannies, the uncles or anyone else in your orbit can build your own adventure book. It’s yours for free ⬇️

There’s a 16+ version in there, too.

Oh, and if you want to use AI for learning, I talked about ChatGPT’s new study mode in last week’s newsletter.

Right. You’ve reached the end of the email. Do you go back to the start and read it again, continue with the rest of your day, or spend a second responding to the poll below?

Blimey, it’s exciting, isn’t it?

See you next week,

Heather and the AIFNT team.

Hi Non-Techies,

The best feedback I can get from my newsletters is when someone says, “y’know what? This is really going to improve my life.” (Or words to that effect. I’m not picky.)

I’m hoping the three things I’m sharing with you today will do exactly that.

Speaking of improving your life (check me out with the segue), a quick reminder that my 3-day ‘Become an AI Trainer’ bootcamp is open. Tickets are selling like hotcakes*.

• Three days of online learning (do it in your PJs?).

• £72/$97.

• Perfect if you’re reskilling (new career) or upskilling (new role).

It’ll take place on 19-21 August 2025, 16:00-17:30 BST / 11:00 - 12:30 ET. Just click the button below to register:

*I’ve literally never purchased a hotcake, and I don’t know anyone who has. Still, apparently, they sell really fast.

Create videos from your documents.

In last week’s newsletter, I name-dropped Google’s NotebookLM like it was a B-list celebrity I used to go to school with.

That was about its audio overview tool, which can turn documents into audio summaries.

Between then and now, they’ve released a video overview tool, which - as you’ve probably guessed by the name - is like the audio summary tool, but it’ll create a video instead.

This (incredibly concise) 37-second video from Google shows what it can do, and our Non-Techie team have tested it out too. The verdict? Whilst the video we created wasn’t totally perfect, it was still super impressive.

This tool is quickly becoming an essential for Non-Techies with a penchant for making life easier.

“Show me the next two scenes”.

Don’t worry, I’ve already Googled it: apparently Aleph is the first letter in the Hebrew alphabet and a mathematical term to represent the “cardinality of infinite sets”. Whatever that means.

It’s also the name of Runway’s new video tool, which they describe as “A new way to edit, transform and generate video.” Bla bla bla. We’ve all heard that before.

But then I watched this 66-second video, and my mind was blown (to be fair, my mind is regularly blown these days - blame AI).

It’s sort of like OpenAI’s Sora, but looks way more powerful and reliable. Like, imagine being able to upload a video and enter a prompt saying “show me this scene from a different angle”. That’s what we’re talking about here.

ChatGPT = your personal tutor.

Ever noticed how ChatGPT is a bit spoon-feedy?

It’s great when you need a clear answer to something, but if you’re trying to learn and retain something new, it doesn’t exactly help.

Well, OpenAI is trying to fix that with study mode, which turns ChatGPT into your own tutor.

I gave it a query using the small amount of maths knowledge I have left in the bank:

Okay, so it’s still pretty spoon-feedy, and this time, it’s literally trying to feed me pi (delicious).

Still, in this newsletter, I chatted about the UK Government’s report on using AI in schools. This feature - whilst probably in need of some refining - seems to be heading in the right direction.

I’m going to spend the rest of the day getting study mode to teach me how to nail a smooth sign-off from these newsletters.

See you next week,

Heather and the AIFNT team.

Hi Non-Techies,

You know when people say, “Sure, AI is smart, but it can’t read your mind.”?

Well…maybe it’ll be able to do that soon, too (at the expense of having to wear a silly shower cap, mind you). Brace yourself: It’s not even the weirdest of the three AI news bits I’m chatting about today.

Before we get going, I’m super pleased to tell you that my third ‘Become an AI Trainer’ 3-day bootcamp is now LIVE!

• Three days of online learning.

• £72/$97.

• Perfect if you’re reskilling (new career) or upskilling (new role).

It’ll take place on 19-21 August 2025, 16:00-17:30 BST / 11:00 - 12:30 ET. Just click the button below to register:

Meet Microsoft’s new…entity?

Do you remember Clippy? That paperclip-with-eyes that used to pop up whilst you were working on a Word doc? Let me jog your memory:

I still have nightmares about Clippy.

Well, it appears Microsoft hasn’t learned its lesson, because its CEO, Mustafa Suleyman, was on a podcast recently and spoke about a new character for Copilot (Microsoft’s flagship AI tool).

Look at this thing:

Apparently, if you look in the mirror and say “Copilot” three times, it appears behind you.

Suleyman elaborated with some slightly unnerving details. “Copilot will certainly have a kind of permanent identity, a presence, and it will have a room that it lives in, and it will age.”

So, if you woke up today thinking, “I wish I had a tamagochi that can help me answer tricky work emails,” you’re in luck. You can read more about it here.

From Prompt to Podcast

There are a few “turn this into a podcast” tools out there, but GenSpark’s latest announcement video really caught my eye.

It’s agentic, which means it can go away and do research around a topic, beefing out your inputs with inputs of its own. You can also choose whether you’d prefer a conversation, a solo-host or a whole group.

How does it differ from Notebook LM?

Whilst Notebook LM offers conversational ‘audio overviews’ of documents, GenSpark is going for proper podcasts. That means voice customisation and top-quality production. Unlike Notebook LM, the aim isn’t to summarise and distil, but to create engaging podcast content around any topic.

I don’t imagine it’ll be long until you can add your own voice to software like this, which will open up a whole load of possibilities for people building a personal brand. For now, it could be a winner for small businesses that want a cheap, easy and fast podcast.

The AI thinking cap.

The scientists are at it again.

Smart people at the University of Sydney are using an AI-powered thinking cap to turn thoughts into text.

The tech is still very much a work in progress (it has a respectable 75% accuracy at the moment), but it could have life-changing implications for people with communication difficulties.

Here’s my Non-Techie explanation of how it works:

• You put on the cap. This part is obligatory.

• You think. The cap reads your brain signals.

• AI interprets those signals and turns them into text on a screen.

It’s demonstrated in this video.

Okay, just like Copilot’s new character, I’m off to go and sit in the room that I live in and age. Don’t forget, there’s a digital seat waiting for you at my next ‘Become an AI Trainer’ bootcamp!

See you next week,

Heather and the AIFNT team.

Hi Non-Techies,

Or should I say, Garr, me hearties.

For the uninitiated, that’s pirate-speak. I’m fluent now, after a wonderful time spent celebrating AIFNT on a pirate ship in Bristol last week.

Yes: there was a drone.

But whilst I was swashbuckling my way around the River Avon, OpenAI were announcing perhaps their biggest launch since ChatGPT itself: ChatGPT agent mode.

Hoist the colours and ready the cannons (the last of today’s pirate references, I promise) - today we’re talking about what agent mode is and whether it’s worth getting excited about.

OpenAI’s Announcement.

When Sam Altman (OpenAI’s CEO) sits down on a sofa and looks down the barrel of the camera, you know something big is coming.

“People wanted a unified agent that could go off, use its own computer, and do real, complex tasks for them”, he explains in the first minute of the announcement video. Okay, that’s pretty big.

How is this different to regular ChatGPT?

ChatGPT is an incredible tool, but right now, it doesn’t take action.

For example, if I want to go on holiday (fat chance), I can ask ChatGPT for trip recommendations and a Spanish language lesson. That’s really cool.

But I can’t ask it to find me an available, affordable hotel for a specific week, create a spreadsheet itinerary, and order a giant, pink flamingo inflatable to my house in time for my departure.

ChatGPT’s agent mode is here to change that. Armed with its own computer and a web browser, it can actually do some of the stuff you’ve asked it.

Agent mode is currently only available to a small number of top-tier users.

Some possible use-cases:

1) Look at my work calendar and create an editable deck I can present in Monday’s meeting using the About page on my website.

2) Plan recipes for a dinner party for next Friday and order the ingredients to my house.

3) Write a whitepaper based on OpenAI’s ChatGPT agent announcement.

Blimey. So far, so mind-blowing. But what do users think?

What’s the initial reaction?

Very few ChatGPT users have access to this, but the headline from early testers is “it’s full of promise, but it needs improvement”. Classic AI rollout, then.

Matt Wolf put it to work in this video, and it didn’t exactly pass with flying colours (that’s not a pirate reference, just pure coincidence).

For example, it took 41 minutes (an eternity in AI terms) to create a pretty mediocre and error-strewn deck.

Even so, Matt acknowledges the massive potential of this tool.

This great LinkedIn post by Allie K. Miller shares a similar sentiment. Sure, ChatGPT’s agent mode needs to be polished and improved, but it’s already proving to be helpful for “anything that needs organised data”. As she puts it, we’re finally seeing that “AI isn’t just chat”.

Our take.

We haven’t used this mode yet, but the thing raising our Non-Techie eyebrows isn’t what it can or can’t do at the moment, but what it’ll be able to do eventually.

I sometimes have to remind myself that we’re only three years into this AI journey. If it’s a road trip, we’re still flicking through Spotify playlists and loading Google Maps, trying to figure out the best route through it all.

The internet was released to the general public in 1993. But do you remember what websites looked like three years after that? Here’s Pepsi’s website to remind you:

Who knew looking at a landing page could give you a migraine?

Sure, agent mode will need improvement, but we can get a real sense of what’s to come: accessible AI tools that can successfully carry out complex tasks for us, whilst we get on with other stuff.

Last week, that felt like an if. Now it’s most definitely a when.

Other things to note:

It’s unclear when it’ll be rolled out to the wider ChatGPT public, but we’re talking days, not years. There’s a chance that by the time this newsletter is sent, it’s already available for most of us.

Sam Altman’s final message in the launch video was a warning: just as society has learned to use the internet safely (for the most part), so too will it need to learn to use AI agents safely.

It just so happens that we put a big emphasis on learning how to use AI safely and responsibly over at my AI Academy. Click the button below if you want to know more about why our members love it:

Right, I’ll be back next week. If you can’t make it, please send your AI agent in your place.

Heather

Hi Non-Techies,

I’ve got some fun updates from AI land today. AI land isn’t an actual place, by the way, but I can see it being some weird, dystopian theme park 10 years from now. Let’s all make a note to organise a group trip once it’s open.

I’m a big advocate for AI (in case you hadn’t guessed), but by pure coincidence, most of this week’s newsletter is about the stuff AI can’t do.

I hope that it gives you a few case studies to whip out the next time someone says that we’re all being replaced.

Claude’s weird, hilarious vending machine challenge.

Regular readers of this newsletter will know that I rate Anthropic’s Claude AI. It’s a great writer, but as with any chatbot, it can do way more than just write.

Anthropic, keen to demonstrate Claude’s abilities in other areas, recently ran an experiment. The premise was simple: Give Claude full responsibility over an office vending machine and see if it can turn a profit.

They created an agent called Claudius, armed it with access to a web browser and gave it an email address that customers could use to request new stock items.

The experiment got pretty weird, pretty quickly. From stocking tungsten cubes to telling its customers it would start delivering products in person (wearing a blue blazer and a red tie, obviously), Claudius went a bit off the rails.

Claude didn’t exactly smash this challenge.

If you fancy a laugh, you can read the whole article here.

Add vending machine operator to the list of jobs that are safe from AI for now.

The AI Fixer.

Speaking of AI taking our jobs, I stumbled upon an article the other week that spotlighted a new job category it’s created: AI Fixers.

Sarah Skidd, a copywriter, is being paid hefty amounts to improve low-quality, AI-generated copy. And she’s not the only one. According to the article, more and more copywriters are finding a niche in fixing AI-written copy that’s either boring, generic or downright wrong.

It’s easy to imagine equivalent roles in other niches. From programming to design to customer service, human involvement isn’t just a nice-to-have; it’s an “OMG we need this ASAP” for businesses using AI to cut corners.

AI Fixers are among plenty of jobs that AI has created. Another? AI Trainers. If you want to save companies from ever needing an AI Fixer, I run a course that’ll transform you from Non-Techie to qualified AI Trainer by mid-October.

Click the button below to learn more about that:

OpenAI’s quiet project.

Like most tech CEOs, OpenAI’s Sam Altman wants his tech to be so deeply embedded in the day-to-day lives of its users that they can’t function without it. Morning coffee, 11am snack, OpenAI.

That might explain why OpenAI has been quietly building a competitor to Google Workspace and Microsoft Office, which will include collaborative document editing and integrated chat features.

I’m intrigued by this, not least because Microsoft owns 49% of OpenAI’s for-profit arm. But it also begs the question, how different can it be? After all, MS Office and Google Workspace already integrate generative AI into their tools.

Will OpenAI’s version really be so much better that it’ll justify the pain of migrating away from the incumbents?

Obvs I’ll be answering that question when it’s rolled out.

I'm giving away something I literally never do anymore.

People constantly ask for 1:1 consultancy sessions with me, but I've been too swamped to offer them since my business grew. However, you now have the opportunity to win one for free find details below!

Hubspot is running "The New Era of Demand: 5 AI-Powered Quick Wins for Modern Marketers" Masterclass with Emmy Jonassen (VP of Demand Generation) and Jonathan Hunt (VP of Media) from HubSpot.

Since we're talking quick wins, here are 3 ChatGPT hacks you can use right now:

1. Stop the AI-speak: Add "Avoid using em dashes and [whatever else annoys you] in responses, or I'll penalise you $200" to your Custom Instructions

2. Make it critique itself: After any response, ask "Is that the best you can do? Identify 3 flaws and rewrite to fix them"

3. Use Projects properly: They're not just folders - they're mini workspaces with separate Custom Instructions for different clients

The masterclass is Thursday 17th July at 3pm BST. Live only, no recording.

In recognition of this masterclass, I’m giving away a free 1:1 consultancy session.

To enter, simply register and attend the live session, then share your top takeaway in a separate LinkedIn post and tag me. I’ll be picking the best one by Friday 18th July 4pm BST  to win a 1:1 session! .

This giveaway is not affiliated with LinkedIn or Hubspot

Register here: https://hubs.la/Q03tNX9M0

Anyway, this Friday I’m captain on a giant pirate ship so I’d better prepare. I’ll explain why in next week’s email.

See you soon, maties (I can’t pull that off, but it’s done now).

Captain Heather

Hi Non-Techies,

I know I promised you food pics from my NYC trip, but then the food arrived, and it looked so good, and I ate it. This just kept on happening, time and time again. I was completely powerless over it.

You’ll have to settle for a pic of me being grabbed by King Kong instead. Don’t worry, I fought him off:

Not even a giant ape can stop these weekly newsletters.

Today, I’m sharing three AI things that I’ve stumbled upon over the last week that made me think of you.

The AI band making waves.

Last week, The Velvet Sundown were accused of being AI-generated.

If you’re not one of their million(!) monthly Spotify listeners, let me fill you in: The Velvet Sundown are a folky band that sings about dust, echoes, silence and driftwood.

But something felt a bit off about them. Nobody knew who they were. They managed to release two excellently produced albums in under a month. Plus, their band pics all looked a bit too…polished.

Here’s one of their songs, Dust on the Wind. The comments are pretty funny:

Earlier this week, they confirmed what many already suspected: it’s an AI project. Here’s what their Spotify profile says now:

“The Velvet Sundown is a synthetic music project guided by human creative direction, and composed, voiced, and visualized with the support of artificial intelligence.

This isn’t a trick - it’s a mirror. An ongoing artistic provocation designed to challenge the boundaries of authorship, identity, and the future of music itself in the age of AI.”

What are your thoughts? Is this “synthetic music project” an exciting new chapter in this mad AI book we’re all flicking through? Or an unwelcome guest sitting at the human-only table?

Copyright your face.

Denmark has proposed an interesting solution to the problem of AI-generated deepfakes (fake media, like images, videos or audio recordings, that pretend to be real).

The Danish government wants to change their copyright law so that everyone owns the rights to their face, body and voice. That way, if someone is the victim of a deepfake, they have a legal basis from which to take action.

It hasn’t been approved yet, but it’s apparently very popular among members of parliament and is expected to be passed. You can read the whole article by clicking the button below:

Did AI write this bit?

Grammarly has released a new, free AI-detecting tool.

AI detectors are nothing new, but Grammarly’s works a bit differently. It’s not designed for the reader; it’s designed for the writer.

Let’s imagine I’m a student writing an essay (thankfully, those days are behind me). My professor has explicitly told me that I can’t use AI. Harsh, but whatever.

Now I can ask Grammarly’s new AI tool to track my document and produce a report that verifies that the words in the document have been typed out, not copied and pasted.

When I submit my essay, I can also attach the report, so that my professor has peace of mind that my terrible mark wasn’t aided by AI, but was instead 100% a result of my own brain.

Incidentally, Grammarly also has an AI detector in the more traditional sense: you paste text into it and it’ll tell you whether AI has written it.

In a very meta move, I pasted this section into it.

Here’s what it says:

I added a hyphen to ‘AI-detecting’ after this…because Grammarly told me to.

Nailed it!

By the way, we tested Grammarly’s new tool in my last weekly AI Academy live session.

If you want immediate access to a friendly (and very human) community, live AI demos and a library of learning resources and prompts, click the button below and join for as little as £25/pm.

Okay, I’m back next week to talk about the time Claude became a terrible business owner.

See you soon,

Heather

Hi Non-Techies,

Is New York really a concrete jungle where dreams are made of, or was Alicia Keys lying to us all? I’ll find out soon, because I’m heading there this week for a holiday.

Consider this your advanced warning that next week’s newsletter is likely to feature at least one picture of a cronut, pizza, or bagel. Possibly all three.

But before I take a bite from the Big Apple, I’ve got a few big bits of AI news to share and discuss. It’s no exaggeration to say that the fuzzy future of AI got a tiny bit clearer last week.

Is AI breaking copyright?

In 2024, three authors (Andrea Bartz, Charles Graeber and Kirk Wallace Johnson) accused Anthropic of stealing their work to train their AI tool, Claude.

Last week, a U.S. judge ruled that using books to train AI tools does not violate U.S. copyright law.

This is major enough that I’d salute it if it walked into the room. Sure, it isn’t the sort of bombshell that’s going to immediately impact your week, but it’s big for the future of AI and how it’s trained.

Context: The AI industry in the U.S. is a bit of a magnet for lawsuits. In the last month alone, Disney and Universal have filed a lawsuit against image-generation tool Midjourney, and the BBC is considering legal action against AI search engine Perplexity.

Regardless of your moral stance on this, the fact that an actual judge has weighed in on the side of AI here might be a sign of things to come for other, similar lawsuits. Here’s what the judge said:

“Like any reader aspiring to be a writer, Anthropic's LLMs trained upon works, not to race ahead and replicate or supplant them — but to turn a hard corner and create something different.”

He hasn’t dismissed the case, though, which means it’ll still go to court. AI for Non-Techies will have its nose pressed firmly against the glass for that one.

“The biggest risk is doing nothing”.

Back to the UK’s seagull-patrolled shores. The UK government has published a report about how the use of AI in schools can impact education, based on 21 early adopters.

Whilst the report itself is pretty unexplosive (the overarching conclusion is essentially, “we need to do more research, but it’s worth exploring”), it’s an encouraging sign that the UK Government is open to exploring the potential impact of AI in education. As they say in the intro:

“The UK government is ambitious for AI and views it as a fundamental part of its mission to break down barriers to opportunity for children and young people.”

But the report also says that stuff like hallucinations and potential safeguarding issues give rise to “an urgent need to assess whether intended benefits outweigh any potential risks”.

AI champions.

There was one particular part of the UK government’s report that really caught my eye:

“Most schools and colleges had an AI champion who was instrumental in getting senior leaders to embrace AI and bringing staff on board…AI champions typically created a ‘buzz’ around AI and played a vital role in demystifying it so that staff began to understand what it was and how they could use it.”

If I could summarise my aim with the AI Academy, it’s to create AI champions everywhere. If you want to be the person who can make AI more accessible for everyone around you, click here to learn more.

See you next week for pizza pics (and maybe some stuff about AI, if time allows).

Heather

Hi Non-Techies,

The Cornetto I’m currently eating is melting quickly in this heat, which means I’m racing to prevent rogue droplets of ice cream from reaching my fingers.

Sound the metaphor alarm, because there’s one coming: Eating this Cornetto feels a bit like keeping on top of AI news. It moves so quickly that, just when you think you’re up to speed, another huge update drops.

To keep our AI Academy members in the loop (or to help them avoid Cornetto-covered fingers, if you’re still hung up on the metaphor), we host a live session every Monday and send a weekly member-only newsletter, too.

I’m trying a similar thing with today’s newsletter. You’ll hear about three big AI news items that have unfolded over the last week. If you enjoy this email, you’ll definitely enjoy being an AI Academy member. Just click the button below to learn more:

Which jobs are the safest?

AI doesn’t have an actual Godfather, but if it did, it’d probably be Geoffrey Hinton (sorry Geoffrey, but that means you owe AI at least a few birthday presents).

That’s because of his pioneering work on neural networks, which is the Techie way to describe how AI learns and improves.

Last week, he was on Steven Bartlett’s Diary of a CEO podcast, where he was asked about which jobs are the safest from being replaced by AI.

“I'd say it's going to be a long time before it's good at physical manipulation…so a good bet would be a plumber.”

And the jobs most at risk?

“For mundane intellectual labour, AI is just going to replace everybody.” He picked out call centre workers and paralegals, in particular.

ChatGPT’s secret notetaker.

ChatGPT released a notetaker last week, but only on the desktop app. It doesn’t appear in meetings (which is a BIG privacy red flag, in my opinion) and drops the meeting notes in a ChatGPT Canvas (that’s the interface you get when you're working on something more complex).

This is a ChatGPT Canvas. You’ve probs seen it before.

I haven’t tested it yet, but Olivia Moore did. She shared her findings in this X thread.

To summarise, it’s not great...yet. It doesn’t recognise names, the notes weren’t that good, and it doesn’t autojoin meetings. No doubt it’ll improve over the next few months, though.

A celebrity endorsement for AIFNT?

I’ve always thought that Brad Pitt would love this newsletter. Now, using AI, I could probably make a video of him wearing an AIFNT t-shirt.

That’s using a tool called Higgsfield.ai, which is causing a proper buzz amongst AI geeks this week. They released a new feature that lets you easily combine your product with an image of a celebrity to create a video that looks scarily real:

I’m more intrigued by the lettuce, to be honest.

Someone grab the tin opener, because this is an ethical and legal can of worms. Still, it’s no surprise that early users are so impressed by the tech itself.

Should I do these update-style newsletters more often? You’d be doing me a huge favour by responding to this poll and letting me know:

See you next week,

Heather