First, my AI PSA:

Well, well, well: we're now 60% sold out and I'm starting to feel a bit smug about who's been paying attention. My 12-week GenAI Productivity Upgrade starts July 14 and frankly, I'm getting tired of watching people get sub-standard results with how they are using GenAI.

Our program will show you the cheeky little secrets, the workflow hacks people don't want to share, and the AI tricks that'll make you look like a bloody genius — but with less than half the spots left and enrolments closing July 9, dithering is becoming rather expensive.

​Snag your spot before we get properly sold out because there's nothing quite like the sting of watching your mates breeze through work whilst you're still doing everything the hard way.

Now, on with today's show:

Do you feel like you're in a gunfight every time you use AI tools? Sending off prompt after prompt, hoping something sticks, then spending ages trying to wrestle the response into something usable?

I've been there too. Desperately trying to get ChatGPT to produce exactly what I need, only to end up with what feels like having a passive-aggressive assistant who deliberately misinterprets every request. "You wanted a professional email template? Here's a limerick about frogs wearing business suits. You're welcome."

Recently, Inventium's GenAI wiz Neo Aplin introduced me to a concept that's transformed how I approach AI: the difference between Gunslinging and Architecting.

The Gunslinger approach

Most of us are "gunslingers" when it comes to AI. We fire off prompts without much planning, hope for the best, and then get frustrated when the results aren't quite right. I think of it like the digital equivalent of throwing spaghetti at the wall and hoping something sticks.

It goes something like this:

1. You ask ChatGPT to write something
2. You get a response that's... okay-ish (worthy of a participation award)
3. You ask it to fix various aspects
4. Repeat step 3 about twelve times (while your coffee goes cold)
5. Eventually settle for something that's "close enough" (and question why you even used AI for the task in the first place).

Sound familiar? This approach might work for simple tasks, but it's inefficient and often produces mediocre results for anything complex. It's like trying to build a house by hammering away without a blueprint – you might end up with shelter, but probably not the dream home you envisioned.

The Architect approach

Architecting with AI is a completely different mindset. Instead of immediately asking the AI to produce something, you first collaborate on the structure and approach – less shotgun wedding, more thoughtful partnership.

Here's what architecting looks like:

1. Establish your goals and constraints upfront
2. Agree with the AI on the structure before any content is created
3. Get the AI to challenge your thinking or play devil's advocate
4. Collaborate on a plan, then execute it together
5. Iterate based on the framework you've established

It's like sitting down with an architect before building a house: "How many rooms do we need? What's the purpose of each space? What's our budget? And can we please avoid that 1970s avocado bathroom suite my mother-in-law keeps recommending?"

When to switch from Gunslinger to Architect

I'm obviously not going to architect a simple email. That would be like hiring a wedding planner to organise your trip to the corner shop. But for complex tasks, architecting is transformative:

• Creating strategic plans or position papers
• Designing workshops or team off-sites
• Developing content frameworks
• Analysing complex data or situations
• Writing anything requiring deep structure

The bigger and more important the task, the more an architecting approach pays off. The time you invest upfront saves hours of revision later – and prevents those moments where you stare blankly at your screen, wondering if your AI has secretly developed a vendetta against you.

Why Architecting works better

When you architect with AI, several things change:

You set clear expectations. By agreeing on goals and structure first, the AI understands what you're really after (instead of playing an endless game of digital charades).

You leverage your expertise. Your knowledge of the context and requirements guides the AI, rather than hoping it can read your mind (spoiler alert: it can't, despite what the conspiracy websites say).

You create a feedback framework. With an agreed structure, feedback becomes about refining specific elements rather than starting over – less "burn it all down" and more "let's tweak this bit."

The AI becomes a genuine thought partner. Instead of just being a text generator, it can help you think through problems and explore possibilities – like having a brainstorming buddy who never needs coffee breaks.

This approach works with any AI model, though the more advanced reasoning models like GPT-o3 are particularly powerful partners for architecting.

Start Small, Think Big

You don't need to completely transform your approach overnight. Try architecting your next substantial piece of work with AI. Start by discussing goals, audience, structure, and tone before asking for any content.

The difference might surprise you – not just in the quality of what you produce, but in how much more satisfying the process becomes.

Happy architecting!

Cheers

Amantha

​

P.S. I'd love to hear if you've tried architecting with AI, or if you have questions about this approach. Hit reply and let me know your thoughts.

​