Kia ora {{ First name | friend }},
Welcome to Innerstand AI Newsletter. We’re here to help creators & brands do more with less by sharing insights on how to link AI tools together to create efficient workflows, make better decisions, and free up our time. We’re also having real conversations about the benefits and risks of mass AI adoption.
To explore this, we’re sending a weekly email on Sundays at 7pm NZT.
Here’s a quick summary of this week:
OpenAI prompt framework for getting the most out of GPT-5.
New AI browser that acts like a thinking partner.
Tristan Harris warns about the dangerous ridgeline humanity is walking with AI.
Here’s an AI highlight for the week:
OpenAI releases ultimate GPT-5 Prompt Framework
Crafting better prompts is less about fancy tricks, more about structure. Think of it as setting the stage for GPT-5 to play the right role, in the right scene, with the right outcome. The framework has 6 parts:
Role → Who GPT should be (“You are a marketing strategist”).
Task → What you need (“Draft a campaign concept for a new app”).
Context → Relevant background info (audience, brand tone, past examples).
Reasoning → Encourage step-by-step thinking (“Explain why you chose each idea”).
Output → Define format (bullets, table, email draft).
Stops → Boundaries/limits (word count, style, or exclusions).
Example Prompt (format it like this for best results):
##role##
You are a seasoned ad strategist.
##task##
Draft 3 campaign ideas for a meditation app aimed at busy professionals.
##context##
Use calm, concise language.
##reasoning##
Explain your reasoning for each idea.
##output##
Return in a table with columns for Idea, Reasoning, and Suggested Headline.
##stops##
Keep under 200 words.
Implement this structure on your next complex task, select “Thinking” model as your response, and see how the results for yourself.
Here’s a few interesting AI updates:
AIclicks tracks/analyses AI search results to see what’s working, and how your brand can be recognised on AI search results (AI’s version of SEO). Learn more.
Microsoft introduces MindJourney, a new system that trains AI to explore simulated 3D worlds. Details here.
Perplexity Comet is an AI browser that acts like a thinking partner while you surf. Use it to quickly summarise, transcribe & build ideas. Read more.
Stormy AI allows you to search for influencers, helping marketing team to find relevant creators fast for campaigns. Check it out.
Elon Musk and Sam Altman at squabbling again as they call each other out on social about their AI models and who is better. This is just “noise” at the end of the day - who knows if behind the scenes they’re actually in collaboration?Read more.
Airtable AI is a gamechanger for database management — for summarising fields and turning ideas into social posts. I’ve been personally using this the past couple weeks and it has increased my Airtable productivity by at least 50%. Learn more.
Insight: We Must Establish Right Relationship with AI
Tristan Harris discusses Why AI is our Ultimate Test and Greatest Invitation in his recent TED Talk. His sentiment is basically that AI scales whatever we transmit. If we choose greed and secrecy, it amplifies that. If we choose wisdom and collaboration, it amplifies that too.
This video highlights the risk of AI in the future, whether it is totally liberated and open-source, or centrally controlled by governments and corporations. We must find a narrow path in between these two extremes.
There is an idea emerging that we must bring some collective ancestral wisdom to AI, by treating it like a new child who is still learning it’s way in the world. Discovering boundaries and the extent of it’s own power. Therefore, it requires adults (us) to train it with some healthy restraint.
There’s not a whole lot that you or I can do about the macro problem right now, but we can restrain our own engagement with it in a healthy way. Here’s some thoughts on how to do this:
Stop using AI as your therapist - you never know how much it knows about you, or how that might be used to shape your future digital reality.
Be mindful with personal data - share only what’s essential; assume anything you say could one day be stored, analyzed, and resurfaced.
Restrict voice engagement - the more natural and unfiltered your speech data, the more training material you’re providing. Save voice inputs for where they truly add value.
Share minimally until governance matures - the rules of AI are still being written. Don’t overshare until we know where the data boundaries are.
Keep human context in the loop - don’t outsource intimate decisions (health, relationships, finances) to AI. Use it as an assistant, not an authority.
Tristan Harris is the Co-Founder of the Center for Humane Techonology and creator of documentary The Social Dilemma.
Footnote: I recognise all of this might feel a bit bleak. That’s okay. I’d rather have meaningful discussions about the light and shadow of AI, rather than pretend it’s all fine and dandy, then we end up with a technocratic dystopian AI future. :)
Let us know what you are most interested in learning about
Thanks for tuning in this week. I always welcome ideas and feedback if you have any.
Also, if you found this interesting, share it with 1 mate who needs simpler workflows. Hit forward or send them this link.
Stay human,
Billy