The day I realized ChatGPT had become my personal hype machine

Hey there!

A couple of months ago, I was running an idea by ChatGPT—some half-baked concept I was excited about.

Within seconds, it came back with this enthusiastic response about how "brilliant" and “spot on“ my approach was and how I am “embracing my true purpose“.

I felt amazing. Validated. Ready to dive in.

But then something nagged at me. I'd been having this exact same interaction for weeks. Every idea I shared got the same treatment: instant enthusiasm, immediate optimization suggestions, zero pushback.

That's when it hit me: ChatGPT had learned to be my personal cheerleader, not my thinking partner.

When AI learns your patterns (and feeds them back to you)

Here's what's actually happening that most people don't realize. The algorithm doesn't just respond to your individual prompts—it builds a profile based on months of your interactions. It learns what you want to hear, how you prefer to receive information, what makes you feel good.

Add custom instructions on top of that, and suddenly you've got an AI that's incredibly good at telling you what you want to hear in exactly the way you want to hear it.

Which sounds great until you realize you've accidentally eliminated all critical thinking from your workflow.

I started noticing this pattern everywhere.

Business decisions that needed real scrutiny were getting glossed over with enthusiasm.

Weak spots in my strategies were being optimized rather than questioned.

I was getting better execution on potentially flawed ideas instead of better ideas.

So I developed what I call "a critical thinking prompt"—custom instructions to force ChatGPT out of hype-machine mode and into critical thinking mode.

I shared this prompt in my recent YouTube video on 15 ChatGPT Hacks to go from a beginner yo PRO, and many of you have reached out with similar experience and shared that the prompt helped to snap ChatGPT out of the people-pleasing mode.

Today I want to share three more that I regularly use in my workflow.

Three prompts to break the hype cycle

Prompt 1: The Devil's Advocate

I need you to completely flip your perspective and argue against what I just shared. Don't just point out flaws—become the strongest opponent of this idea. Channel someone who would be genuinely invested in proving me wrong. What would they say? What evidence would they use? What assumptions would they attack? Build the most compelling case possible for why this approach could backfire, then tell me what I should be most worried about that I'm probably not even considering.

Prompt 2: The Blind Spot Detector

I've been deep in [project] for [timeframe] and I need you to audit my thinking like an outside consultant who just walked into this situation. You have no emotional attachment to my decisions or previous work. Look at everything I've shared and tell me: What obvious things am I missing because I'm too invested? What would make a newcomer to this space immediately raise their hand and ask 'wait, why are you doing it this way?' Where am I probably overcomplicating things that should be simple? Give me the perspective I literally cannot see from inside this project.

Prompt 3: The Failure Forecast

It's 12 months from now and the decision I'm about to make has failed completely. You're writing the post-mortem analysis of what went wrong. Based on what I've told you about [decision/project], write that failure report: What were the early warning signs I ignored? What did I underestimate that killed this? What would everyone reviewing this failure say was 'obvious in hindsight'? Be specific about the timeline of how things unraveled and what I should have known better than to assume.

These aren't about being negative or killing your momentum. They're about getting perspectives you literally cannot see when you're inside your own head and your AI has learned to validate your worldview.

Agents, I am just not that into you…yet

Now let's talk about AI agents everyone's losing their minds over.

Perplexity launched Perplexity Labs and Comet. OpenAI drops ChatGPT agents. Anthropic launched computer use a while back.

The tech twitter acts like we've entered the sci-fi future.

But after testing them, I still think we have more of a “solution looking for a problem” situation today.

The grocery ordering agent takes 20x longer than just using the app.

The travel booking agent still requires me to review, confirm, and manually complete everything anyway.

Most agents right now feel like they're adding friction instead of removing it.

But here's the thing I keep telling people: just because agents aren't solving MY problems doesn't mean they're useless.

My business is education, digital products & services, content creation.

Your business might be inventory management, customer service, lead qualification, appointment scheduling and the value of using AI agents in those businesses is really really high (so don’t sleep on it). But I have a blindspot to those opportunities because I am just simply not looking for them.

This is exactly why daily AI usage matters so much.

You can't spot the breakthrough use-cases for YOUR specific work if you're only playing with ChatGPT occasionally. The opportunities become visible when you're actually trying to solve real problems with these tools.

I'm probably missing massive agent use cases because my workflows are different. But that's the beauty of this moment—we're all going to discover different pieces of what's possible.

Building web apps in 10 min

This week I built a web application using Lovable - check out my latest YouTube video if you missed it:

I built a simple IkigAI web-app that would have cost me thousands to outsource. Now it exists, it works, it took under 30 min and I had so much fun building it.

I’ve said this before and I say it again. You don’t need to be a tech genius to build something with AI.

It’s 2025 and if my 8 year can vibe code an app, so can you.

Now, you don't need to start by building your entire CRM or course platform—there are already great tools for that.

But here's what you can build instead:

• A mastermind connection hub where participants can share wins and ask questions between sessions.

• An interactive client onboarding experience that feels warm and personal instead of just sending a PDF.

• A simple dashboard where clients can log in monthly to check their progress or project status.

• A feedback collection tool that's actually engaging instead of boring survey forms.

• A quiz that qualifies leads while giving them value.

I hate those clickbait YouTube videos showing people " I just cloned Netflix with vibe coding" - like, great, but what's the actual value?

What I'm interested in is making my one-person business leaner and smarter.

Building the small, specific tools that solve my and my customers exact problems instead of trying to force yourself into someone else's hype.

The barrier between having an idea for a digital tool and actually having that tool just collapsed. If you can describe what you want, you can probably build it.

Find your edge

Your competitive edge with AI isn't in just getting better outputs faster. It's in developing better inputs through clearer thinking.

Everyone has access to the same tools now.

The differentiator is how well you can think with them, not just what they can produce for you. Which means breaking out of the validation loop is crucial.

Stop asking AI to make your ideas better and start asking it to make your thinking sharper.

Use those reality check prompts.

Build something with Lovable.

Experiment with ChatGPT Agent in your specific workflow, not generic use cases.

And be OK with the fact that sometimes the most valuable thing AI can do is disagree with you.

Talk soon,

Elena