Ideally, a walk is just for walking.

But I have a lot to get through in a day.

So I imagined myself strolling through the neighbourhood, a typewriter strapped to my chest, desperately trying to tap out my thoughts while dodging trees and concerned neighbours.

I then realised I'm not ready for this future.

And neither are my neighbours.

But still, I thought – to quote the Beach Boys – wouldn't it be nice, on those busy days, to get some writing done and some movement in at the same time?

The Walking-Thinking Connection

Walking has always been the secret to deep thinking.

Away from screens and notifications, the mind wanders freely, making connections that stay stubbornly hidden when you're hunched over a keyboard.

For years, I've had my best ideas during walks.

An Unexpected Breakthrough

But what if we could not just think but write while walking? Mind you: not with a chest-mounted typewriter but through something far more elegant: AI-powered dictation.

This morning, I discovered that Claude.ai's newly updated dictation feature has quietly transformed the walking-writing equation.

I've been supremely frustrated with Apple's native dictation, and it's probably going to take years before it will be as good as OpenAI's Whisper dictation platform in the EU, what with the AI Act and all.

But dictating to Claude.ai, the accuracy and natural flow are remarkable enough that I wrote this entire piece while walking through my neighbourhood. It wasn’t perfect. But neither is sitting at my desk all day.

Why This Matters for You

Think about your typical workday. How many hours do you spend sitting, trying to force creativity while staring at your screen? What if you could:

• Draft emails during your morning walk

• Outline presentations while getting fresh air

• Write reports while moving through nature

• Stay healthy while staying productive

The Bigger Picture

Walking while working fundamentally transforms our relationship with productivity.

AI tools now offer an unprecedented opportunity to break free from our desks while maintaining output.

Here we find the sweet spot: technology enhancing our natural thinking processes rather than constraining them.

Your Turn

I'm curious: Have you tried combining movement with work?

What's holding you back?

Share your thoughts or experiences – I'll be reading your responses during tomorrow's walk.

Written entirely while walking, with an AI assistant considerably lighter than a typewriter.

AI Signal of the Week: Red Panda's Identity Revealed

The mysterious "Red Panda" AI image generator that suddenly jumped to the top of an AI leaderboard has been unveiled as Re-craft.ai, a newcomer specialising in handling text-heavy images - a persistent pain point in AI image generation.

While Re-craft.ai generates more readable text than Midjourney, it currently trades off some of Midjourney's stunning visual quality to achieve this text clarity.

I tested both this week and included comparison shots below - as you can see, Recraft can handle longer text but my conclusion is that Midjourney still blows other image generators out of the water in terms of image quality.

Text generation just isn't there yet. Of course there is also Ideogram, but it only does short text well, in my experience.

The text-in-image challenge remains one of AI's most frustrating unsolved problems - despite billions in investment and years of development, even the best models still struggle with long-form text generation in images, suggesting this could be the next major breakthrough area in AI imaging.

Why This Matters For You: Text-heavy visuals (presentations, infographics, social media cards) are the backbone of business communication - whoever cracks the text-in-image problem will transform how we create visual business content. (If you're using Midjourney, don't miss this week's Power Move tutorial below for advanced insights on prompting techniques)

Future Focus: Computer-Using AI Has Arrived

Remember teaching your parents how to use a computer? ("No, the other mouse button!") Well, AI just enrolled in the same class – and the implications are massive.

Anthropic's Claude can now use computers like we do: clicking buttons, typing text, and navigating screens. Think of it as giving AI an actual driver's license for your computer instead of just being a backseat driver.

Some experts are sounding alarm bells. Web developer Morten Rand-Hendriksen warns this could enable sophisticated scams – imagine AI impersonating your voice while clicking through websites. His stance? Just because we can build it doesn't mean we should.

But here's the thing: This capability represents exactly what many people have been waiting for – AI that can handle those mind-numbing sequences of clicks and keystrokes that eat up our workdays.

While the current version is reportedly "slow and error-prone," remember what AI experts always say: This is as bad as the technology will ever be. It will only get better from here.

Power Take: Whether we embrace it or resist it, AI is stepping out of the chat window and into our actual workflow. The question isn't if this will transform how we work - that’s for sure. The real question is how we'll adapt to work alongside it.

The Lazy Expert's Guide to Perfect Midjourney Prompts

While many guides advocate for complex prompt engineering, my recent experiments with Midjourney have revealed a surprising truth: simpler prompts often produce more aesthetically pleasing results. After extensive use and training Midjourney to understand my personal aesthetic preferences, I've discovered that less is definitely more.

Training Midjourney on Your Style

Before we dive in, it's worth noting that my results are influenced by consistently using Midjourney and developing what I consider my "aesthetic signature."

The more you use Midjourney to rank images side by side, the more it learns your style preferences. This means that even simple prompts can produce highly personalised results that align with your visual taste.

The “Traditional” Framework

Many Midjourney guides recommend this structured approach:

[Subject] with [Key Attributes], in [Style and Mood], featuring [Colors and Lighting], including [Optional Elements]. [Exclusions]

Because writing such detailed prompts can be tedious, many users turn to AI assistants like Claude.ai to generate these complex prompts. While this makes the process easier, it often leads to overcomplicated results that lose their conceptual clarity.

The Problem with AI-Generated Complexity

Let's look at some prompts typically generated by AI assistants:

Sound Visualisation

Compare the output of the first complex prompt with the second more simple one

Another example: Neural Networks

Compare the complexity above with the simplicity below

Last Example: Dance Movement

Again, first complex, then simple.

The image above is not bad, but not as compelling as the one below:

I don’t know about you, but for me all this clearly steers me in the the direction of: “Make your prompts as simple as possible.”

So while many so-called experts advice us to use AI to help us generate elaborate prompts, these often create busy, overcrowded images where the core concept gets lost in technical specifications.

The simpler prompts, however, produce clearer, more aesthetically pleasing results that better capture the essence of what we're trying to visualize.

Why Simple Works Better

1. Conceptual Clarity: Simple prompts focus on the core idea without getting lost in details

2. Aesthetic Freedom: The AI can apply its trained understanding of composition and style without conflicting instructions

3. Better Interpretation: Less conflicting information means clearer artistic direction

4. Visual Impact: Clean, focused images often have more immediate appeal

5. Personal Style: Simple prompts allow your trained aesthetic to shine through

The Power of Personal Style

What makes this simple approach particularly effective is the relationship you develop with Midjourney over time. As you:

• Select and upscale specific variations

• Create new versions of images you like

• Consistently choose certain styles and aesthetics

The AI begins to understand your preferences, making simple prompts increasingly effective at producing results that match your vision.

When to Use Complex Prompts

The structured framework and AI-assisted prompt generation still have their place:

• Technical documentation needs

• Exact brand requirements

• Scientific visualisation

• Specific architectural or product renders

• Situations requiring precise control over multiple elements

But in all these cases, there are many other models that will produce results that will more closely adhere to your prompts, but are (I find) less aesthetic.

The New Approach to Prompting

1. Start with the core concept (2-3 words)

2. Add only essential modifiers if neede

3. Trust the AI's artistic training

4. Iterate based on results, not specifications

5. Consistently upscale and create variations of images that match your aesthetic

6. Let your personal style evolve naturally through consistent use

Building Your Aesthetic

Instead of spending time crafting elaborate prompts, focus on:

• - Identifying what you like in generated images

• Creating variations of successful results

• Developing a consistent style through your choices

• Learning which simple prompt words trigger the best results

• Building a library of effective, simple prompts

The goal is to create compelling visuals, not to write technical specifications.

Sometimes the most powerful prompt is the simplest one, enhanced by your developed aesthetic relationship with the AI. L

Let Midjourney learn your style, and you'll find that less really does become more.