The Hidden Decisions You're Handing Over to AI

Hi friends! Earlier this week I asked Claude to "help me draft a strategy plan for Q1," hit enter, and stared at the result wondering why it felt so generic.

The problem wasn't Claude. The problem was me.

Without realizing it, I'd handed over about 47 micro-decisions to an AI. What perspective should this come from? What format works best? What trade-offs am I willing to make? All of those questions got answered by default settings I never consciously chose.

So naturally, I started experimenting. And what I found changed how I prompt completely. Let's get into it!

What's inside this week:

This week’s newsletter sponsor

ChatGPT is a superpower if you know how to use it correctly.

Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.

Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.

Download the free guide.

Founding Membership Extended Through February!

Before we dive in, some exciting news: I'm extending the Founding Membership through the end of February!

If you haven't joined yet, Founding Member pricing (50% off) is available through February. This is the lowest it will ever be. Lock in your spot here.

If you are a curious reader who would like to attend a workshop this month and actually experience what we're building before committing, this is for you. Our community workshops are hands-on, practical, and focused on solving real problems you're facing right now.

Upcoming Workshops:

Browse the full calendar → Luma Link

When you sign up, select "Guest of AI Snack Club" and come see what we're all about!

🦋 The Invisible Decisions You're Delegating

When you type "build me a landing page" into a tool like Lovable or "reply to this email" into ChatGPT, you're not just asking for help. You're giving AI permission to make a bunch of assumptions about what "good" looks like.

And AI will happily oblige. It'll pick a layout. It'll choose a tone. It'll decide what information matters most. But it's making those calls based on patterns from millions of generic examples, not based on what actually works for your specific situation.

Take that landing page example. When you say "build me a landing page for my course," the AI is silently deciding:

Should the CTA be above the fold or after social proof? (This changes conversion rates by up to 30% depending on your audience's awareness level.)

Should we lead with the problem or the transformation? (B2B audiences often respond better to problem-first; B2C often wants the dream state first.)

How much copy is too much copy? (High-ticket offers typically need longer copy. Impulse purchases want shorter.)

The AI isn't thinking through these trade-offs based on your business model, your price point, or how aware your audience is. It's just picking something that looks reasonable to a robot.

🦋 What I Started Doing Instead

The difference between okay AI output and genuinely useful output is usually just one step: pausing to identify what decisions you're about to hand over, then making those decisions yourself before you prompt.

Here's what this looks like in practice:

For building things (apps, websites, presentations):

🐛 Before: "Build me a dashboard for tracking my clients"

🦋 After: "Research the top dashboard layouts from a UX perspective. Show me the 3 layouts that perform best for retention and quick decision-making, and explain the trade-offs of each. Then help me choose which one fits my use case: I'm a solopreneur who checks this dashboard 2-3 times per day on mobile and needs to spot red flags immediately."

Now I'm making AI do the research work and show me options, rather than letting it silently pick option #1 from its training data.

For strategy and planning:

🐛 Before: "Create a go-to-market strategy for my new feature"

🦋 After: "Develop three versions of this go-to-market strategy from different executive perspectives: one from a VP of Marketing (focused on positioning and channels), one from a VP of Finance (focused on unit economics and payback period), and one from a CEO (focused on strategic timing and competitive positioning). Highlight what each perspective prioritizes and where they conflict."

Now I'm getting multiple strategic lenses instead of one generic plan, and I can see where the trade-offs actually are. (Side note: Watching fictional executives argue with each other is weirdly entertaining.)

For everyday stuff:

🐛 Before: "Give me dinner ideas"

🦋 After: "Give me ideas for cooking dinner tonight from two perspectives: one from a Michelin star chef focused on flavor and technique, and one from a scrappy college student focused on cheap ingredients and minimal effort. Show me how I can make the most delicious meal possible with what I probably already have in my pantry."

Now I've defined the constraints that actually matter to me (taste + budget + convenience) rather than getting random recipes I'll never make.

🦋 The 30-Second Check

Next time you're about to prompt AI for something that matters, try this:

Then rewrite your prompt with that context included. You don't need to be a prompt engineer. You just need to be clear about what you want before you ask for it.

The irony is that this actually makes AI MORE useful, not less. When you're intentional about the parameters, AI can focus its intelligence on the actual hard part (synthesis, ideation, drafting) instead of guessing what you meant.

What’s Hot on the Internet