You open a clean chat window. You type: “Write me a go-to-market strategy for my new SaaS.”
Three seconds later, the cursor starts flying. Bullet points emerge. Bold headers appear. It’s organized. It’s articulate. It looks like work.
You feel a surge of productivity. This is the industrial-grade dopamine hit of the AI era. You’ve just bypassed three hours of staring at a blank Notion page, and in exchange, you have 800 words of “strategy.”
There’s just one problem: it’s useless.
The Dopamine Hit vs. Reality
Completion bias is a hell of a drug. Our brains are wired to equate “finished text” with “solved problem.”
When the AI gives you a plan instantly, your brain marks the task as done. You feel the relief. But you haven’t actually made a decision. You haven’t navigated a trade-off. You haven’t sharpened your thinking.
You’ve simply generated a hallucination that wears a suit.
The Hallucination Hall of Fame
If you don’t give an AI business advisor enough context, it doesn’t stop and ask for more. It guesses. It fills the void with whatever statistical probability dictates comes next.
In the high-stakes world of fractional executives and founders, this leads to what I call “Strategic Slop.” Here are three real-world-ish examples of one-shot failures:
The Ghost City GTM: A 12-page Go-To-Market strategy for a luxury pet-sitting business, complete with demographic analysis, for a city that literally doesn’t exist.
The Influencer Delusion: Confidently recommending a $50k TikTok influencer campaign for a B2B infrastructure startup that only has three potential customers globally, all of whom are 60-year-old CTOs who haven’t downloaded an app since 2014.
The Starbucks Rivalry: Identifying Starbucks as the primary “local rival” for your two-person office coffee service startup, complete with a fake 10-page market share analysis comparing your $500 monthly revenue to a $100 billion global empire.
These aren’t just “bugs.” They are the natural result of asking a machine to solve a problem without giving it the variables.
The ‘Average’ Trap
Standard business strategy tools that rely on one-shot prompting fall into the “Average Trap.”
An LLM is, at its core, a world-class pattern matcher. If you provide a vague prompt, it provides the most likely average response based on the vast sea of generic internet content it was trained on.
It assumes you are a generic version of your industry. It assumes your budget is “standard.” It assumes your risks are “typical.”
But strategy is never about the average. Strategy is about the edge. It’s about the specific, ugly constraints that make your business different. If you aren’t feeding the AI those constraints, you are just asking it to rewrite a mediocre Wikipedia article about your job.
The Problem I Kept Running Into
I started testing AI writing tools in 2021. Jasper was one of the first, originally called Jarvis until Marvel’s lawyers had something to say about it. A hundred dollars a month on AppSumo. I also picked up Nichesss around the same time, another early AppSumo find built for content generation.
Neither one was a miracle. They were pattern matchers with clean interfaces.
But I kept watching. Every tool that followed got measurably better at output. The writing got cleaner. The structure got tighter. The speed got faster.
The core problem never changed.
Every single one of them produces weak output when you give it a weak input. Garbage in, garbage out. That’s not a model problem. That’s a sequence problem. The tool can only work with what you give it.
That’s what drove how I built 95Confident. Several modules don’t just ask what you think. They ask you to back it up. If you tell the strategy module your market is underserved, it asks how you know that. If you say your price point is competitive, it asks you to show your work.
The friction isn’t a bug. It’s the whole point.
Context is the Only Moat
In 2026, the AI output itself is a commodity. Anyone can generate a SWOT analysis in five seconds.
The only thing that still has value, the only moat you have as a consultant or a founder, is context.
Your actual burn rate. Your founder’s specific refusal to use LinkedIn. The fact that your lead engineer is leaving in three weeks. The specific regulatory hurdle in the DACH region that no one talks about.
If your strategic planning software doesn’t know these things, it cannot help you. It can only reflect your own vague assumptions back at you in better prose.
The Socratic Correction: Friction as a Feature
Most AI tools try to remove friction. They want to give you the answer as fast as possible so you keep clicking the button.
We did the opposite. We built friction in.
When you use the Business Strategy or SWOT Analysis modules on 95Confident, you don’t get an answer right away. You get hit with a gauntlet.
Four to seven focused, Socratic questions. One at a time. No skipping. No vague answers accepted.
It’s annoying. It makes you sweat. It forces you to look at your spreadsheets and your calendar. But that friction is the feature.
By the time you reach the end of the questioning phase, the AI isn’t guessing anymore. It’s building a structured brief based on the ugly specifics of your actual situation.
You aren’t getting a prompt. You’re getting a thinking partner that has been forced to understand your moat.
The Verdict
If the AI didn’t make you think before it answered, it didn’t actually help you.
One-shot prompting is for social media captions and recipe ideas. It is not for high-stakes decisions. It is not for $5,000 monthly retainers. It is not for the fractional CXO who needs to walk into a board meeting with a plan that won’t fall apart under the first “Why?”
Stop looking for the shortcut. Start looking for the gauntlet.
Ready to move past generic prompts?
Subscribe to The Brief: Get our weekly deep-dive on strategic thinking and AI at 95Confident.substack.com.
Try the Gauntlet: Run your next big decision through our Socratic engine. No generic outputs. Just structured briefs you can actually use. Start at https://app.95confident.com.
