AI Isn't the Problem—Your Prompts Are

Todd Brooks, Founderupdated July 23, 20264 min read

Tangled white threads resolving into clean red streams between two small silhouettes

In short

Most disappointing AI output is a delegation problem, not a model problem: the request never specified the audience, the constraints, or what a good answer looks like. This lays out the POWER framework for structuring a prompt, and the advanced techniques worth adding once the basics are holding consistently.

Jump to a section8

Transform AI frustration into competitive advantage with the POWER framework and advanced prompting techniques for B2B SaaS founders.

Let's be honest. If you've ever rolled your eyes at an AI response, you're not alone. B2B SaaS founders tell me the same story again and again:

"We tried ChatGPT. We asked for a sales email. It gave us something… fine. But not useful. Not what we'd actually send."

The frustration is real. You're running a business, not an experiment in mediocrity. You need clarity, precision, output that matches the high stakes of your next campaign, investor deck, or product launch.

Here's the truth most AI hype glosses over:

AI isn't broken. Your prompts are.

Why Founders Struggle With AI

Imagine briefing a brilliant but literal-minded intern. They don't know your company. They don't know your customers. They don't know what matters to you unless you tell them. That's AI.

Most founders hand over prompts like:

  • "Write some copy for my website."
  • "Summarize this sales call."
  • "Give me ideas for content."

And then they wonder why the results are generic, shallow, or just plain wrong.

AI is a pattern matcher. It doesn't "understand" your business. It predicts the next likely word based on training data. Unless you feed it specific context, structure, and requirements, it will happily serve you vanilla responses.

So when you feel let down by AI, the real gap isn't the model—it's the prompt.

The POWER Framework

That's why I use (and teach) the POWER framework:

  • P – Purpose: Be crystal clear about your goal.
  • O – Output format: Tell AI how you want the response structured.
  • W – Who: Define your audience and perspective.
  • E – Examples: Show it what good looks like.
  • R – Requirements: Add constraints that sharpen focus.

Let's unpack that.

A bad prompt: "Write a LinkedIn post for my business."

A POWER prompt: "Write a LinkedIn post (Purpose) that positions me as an expert in B2B SaaS pricing (Who), using a Seth Godin–style voice (Examples), structured in 5 short paragraphs (Output), under 200 words and ending with a question (Requirements)."

See the difference? The second one doesn't leave room for AI to guess. It's directed. It's useful. It feels like you.

Advanced Prompting Techniques

Once you've mastered the basics, you can unlock far more powerful results. A few methods I teach:

1. Chain-of-Thought Prompting

Instead of asking AI for the final deliverable, ask it to walk you through the thinking step by step.

Example:
"Help me draft a sales email. First, list the top 3 objections my ICP has. Then outline a persuasive structure. Finally, draft the email."

2. Role-Playing Prompts

Assign AI a persona with relevant expertise.

Example:
"You are a direct response copywriter who has generated $100M in SaaS sales. Draft a landing page headline for a founder-led SaaS targeting mid-market CFOs."

3. Constraint-Based Prompting

Limitations force better output.

Example:
"Write a 150-word sales script without using the words: boost, growth, scale, improve."

4. Comparative Prompting

Ask AI to compare options and explain.

Example:
"Compare these two pricing-page headlines: (1) Save 20% With Smart Automation. (2) Stop Burning Hours on Manual Work. Which resonates more with SaaS founders raising Series A?"

These aren't "AI tricks." They're business tools. They force clarity. They turn "random text generator" into "collaborative strategist."

Why This Matters for SaaS Founders

As a founder, your time is your scarcest resource. You don't have hours to rewrite AI's generic answers.

Prompt engineering changes the ROI equation:

  • Faster drafting of investor updates that actually read like you.
  • Sharper sales collateral tailored to your ICP's pain points.
  • Cleaner customer success scripts that reduce churn.
  • Insightful market research summaries that feel boardroom-ready.

AI won't replace you. But with the right prompts, it will amplify you.

Think of prompt engineering as a leadership skill. It's not about typing clever hacks into ChatGPT. It's about communicating goals so precisely that the output accelerates your strategy.

The Iteration Mindset

Here's another truth: the first prompt is rarely perfect.

The process looks like this:

  • Start with a basic prompt.
  • Evaluate the gaps.
  • Add constraints.
  • Iterate until the output fits your voice and your goals.

This isn't extra work. It's leverage. Each round sharpens the AI into a partner that delivers clarity, not clutter.

From Frustration to Flow

When I coach founders on prompt engineering, I see the same shift:

  • Before: "AI is a toy. It's not serious enough for my business."
  • After: "AI just cut my prep time in half and gave me insights I actually trust."

The difference isn't the tool. It's how you talk to it.

Prompt engineering transforms AI from "nice-to-have" to "competitive edge." And in SaaS—where speed, clarity, and customer trust matter—it's not optional anymore.

Your Next Step

If you're a B2B SaaS founder, here's the uncomfortable truth: Your competitors are already figuring this out. They're not settling for average LLM responses. They're building systems where AI drafts, analyzes, and strategizes at the speed of business.

You don't need another tool. You need better prompts.

So let's make this practical:

👉 Drop your biggest AI frustration in the comments.
Maybe it's sales copy that sounds robotic.
Maybe it's customer emails that miss the nuance.
Maybe it's market insights that feel like fluff.

I'll show you how a single engineered prompt can flip that frustration into clarity.

Frequently Asked Questions

Why do founders get disappointing AI output?

Because the model is a pattern matcher predicting likely text, not a system that understands your company, your customers or what matters to you. Hand it write some copy for my website and it will produce something reasonable for a business it knows nothing about, which is exactly what you asked for.

What is the POWER framework?

Purpose, the goal stated plainly; Output format, how the response should be structured; Who, the audience and perspective; Examples, showing what good looks like; and Requirements, the constraints that sharpen focus. Five slots, filled before you ask.

What does that look like in practice?

The contrast given is write a LinkedIn post for my business versus a prompt naming the positioning, a specific voice to write in, a five-short-paragraph structure, a 200-word ceiling and a closing question. The second leaves the model nothing to guess at.

What is chain-of-thought prompting used for here?

Asking for the reasoning rather than the artefact. Instead of requesting a sales email, you ask for the top three objections your ICP has, then a persuasive structure, then the draft — so you can correct the thinking before it is baked into prose.

Why do constraints improve output?

Because limits force specificity. The example is a 150-word sales script that may not use boost, growth, scale or improve — removing the default vocabulary leaves the model no choice but to say something concrete about the actual product.