Issue #52: Bionic Prompts for alpha, strategy, high-quality outputs

Good morning.

Over the years, I’ve held in-depth workshops on prompts and prompt engineering (the real kind, not the marketer’s version).

Between 2023 and into 2024, I gave talks on what I call Bionic Prompts.

This is just a fun name to a prompting process I developed between 2020 and now. I’m always updating it, as the way we interact with LLMs and their capabilities is changing.

I’ve collected notes on my Bionic Prompts talk and I’m sharing them with you in this issue.

Let’s get into it.

—Sam

The New Reality of AI

AI is no longer an unfair advantage - it's table stakes. The resources that were only available to enterprise companies four years ago are now accessible to everyone. 

Your competitor, even if they're a mom-and-pop shop, has the same AI capabilities you do. This levels the playing field completely.

Everyone can prompt a model—and improve their skill.

So, just knowing how to prompt is not enough for an unfair advantage

As I’ve said many times before:

Your prompts are only as good as the context you give them.

If you provide highly relevant information, you’ll get better responses—even if your prompt is only a few sentences long.

Ignore the silly mega prompts that people are shilling. With GPTs, Projects, and Agents, your system prompt matters more than anything and enables you to have a simple conversation, instead of copy-pasting 2,000 word prompts (which is a terribly flawed way of prompting).

You can find more information on this in these two issues: 

Issue #47: 40+ Prompts for Copywriting and Buying Psychology

Issue #48: Prompts for Messaging, Bullets, and Content

Let’s continue.

The Paradox of Prompts

There's a fascinating paradox when it comes to prompts: they're simultaneously worthless and incredibly valuable. 

Some of the prompts I'll discuss have helped my clients generate significant revenue.

However, it's crucial to understand that prompts don't change fundamental business principles.

If your offer isn't solid, no prompt will help you - in fact, it'll make things worse. The fundamentals of offers and lists are more important now than ever before.

The Truth About Prompt Engineering

Let's address something important: prompt engineering isn't what many marketers claim it is. If you see a sales page from a newly minted "AI expert," be skeptical. 

True prompt engineering involves developers using language and code for symbolic manipulation of large language models - it's far above most marketers' pay grades. We’re talking the use of Python, API calls, and a lot more (even down to requesting specific tokens from an LLM).

What we're discussing today is constructing effective prompts for business and marketing purposes. I’m not going technical with these notes.

Technology on Human Terms

For the first time in perhaps hundreds of years, we have the opportunity to use technology on human terms. 

Previously, we had to adapt to technology's requirements - learning to code, navigating complex platforms, clicking through menus. 

We've contorted and distorted ourselves to fit technology's demands. But now, the switch has flipped.

You can now speak or type in natural language, and things get done. Your business no longer needs to conform to what technology can do - technology can adapt to your needs. 

You don't even need technical knowledge beyond basic human communication. 

Understanding the Cynefin Framework

The Cynefin framework, created by David Snowden (Cynefin is Welsh for "place"), helps us understand where we are in terms of business complexity. 

For the past century, marketing and sales operated in clear or complicated domains where direct response tactics worked because we could analyze, categorize, and expect predictable results.

We're no longer in that world. We've moved from clear/complicated domains into complex/chaotic ones.

The past three years have been chaotic, and we're hopefully moving toward complexity. In these domains, traditional linear thinking doesn't work. 

Instead, we need to:

• Probe and sense what's happening

• Act with limited information

• Get feedback quickly

• Adapt our approach based on results

The time frame for feedback has collapsed dramatically. What used to take weeks or months now happens in minutes. 

You can create something with AI in five minutes, launch it on Facebook, and get feedback in another five minutes. Your feedback loop can be as short as ten minutes.

Process Over Prompts

Let's break down a few typical prompting approaches (there are many more):

1. Zero-shot prompting: Coming in cold with no context, like saying "Write me an email about my blue widget." It's equivalent to walking up to someone at a bar with a cheesy pickup line - the results are unpredictable.

2. Few/Multi-shot prompting: Starting with context and working through a process, adjusting based on feedback.

3. Chain of thought: Asking AI to think through steps systematically. You can literally say "Let's think this through step by step" and it will show its internal process.

4. Tree of thought: This is where the magic happens. You ask for alternatives and evaluations based on specific criteria. For example, asking for email variations scored on persuasiveness (but never allowing a perfect 10/10 score).

The Fundamental Schema

There are four types of sentences in English that form our prompt framework:

• Imperative: "Turn on the water"

• Declarative: "The water is on"

• Interrogative: "What temperature is the water?"

• Exclamative: "The water is too hot!"

Historically, we've interacted with technology through imperative and declarative statements.

Now, we're moving into interrogative and exclamatory territory. This allows for more nuanced communication and better results.

You can apply any of these four sentence types to your prompts and improve the quality of outputs.

The Three Core Components of Effective Prompts

The foundation of effective prompting lies in three key components: function, context, and modifiers. While you don't need all three in every prompt, incorporating them will significantly improve your results. Let's break down each component in detail:

1. Function: This is where you establish the fundamental parameters of what you want to achieve. 

It includes:

• What you want to happen

• Who you want the AI to be

• How you want it to approach the task

• The objective you're aiming for

A crucial note about function: Avoid telling the AI to "act like a direct response copywriter" - this will lead to clichés and tropes. 

Instead, frame it like this: "You're an expert marketer with rigorous, detailed, and experienced understanding of persuasion, influence, marketing, and sales, and can craft messaging that leads people to make a decision."

When writing HR documents, you wouldn't tell it to be a copywriter - instead, you might say "you are precise in your language, you know exactly how to formulate your thoughts and sentences into a coherent message that people love to read."

2. Context: This component provides the necessary background and parameters for the task. 

Include:

• Relevant data and information

• Keywords and key phrases (not in the SEO sense, but terms you want to include)

• Avatar and audience details

• Product information

• Competitor insights

• Examples (like "Here's a good ad, do an ad like this but for this product instead")

• Market awareness categories

• Any relevant trends

Context doesn't need to be exhaustively long - even 2-3 sentences can be sufficient if they contain the right information.

3. Modifiers: These are the specific instructions that shape the output's style and tone. Think of adjectives as your best friends here.

 Include:

• Descriptive language preferences

• Mood and voice specifications

• Style requirements

• Syntax preferences

• Point of view

• Structure guidelines

• Emotional intensity ("amplifying increased emotion by 200%")

• Language style ("spoken like a human would with emotional descriptive fascinating language")

Important notes:

• These components don't need to appear in strict order

• One paragraph can contain elements of function, context, and modifiers

• You can split them up or combine them as needed

• The process is generative, not just iterative - what the AI tells you should influence how you adjust your next prompt

• Think of it as giving instructions to a smart, talented, eager assistant who's waiting to help you

• You can mix instructions across these categories - there's no need for rigid separation

Remember: Bad communication equals bad prompts. If you know how to talk to another human being and get what you want out of that conversation (not in a greedy way, but with a clear outcome in mind), you'll be able to craft effective prompts.

Examples of Prompts

Remember, these prompts assume a larger context available via knowledge files and data sources, usually via GPTs, Projects, or Bots—even Agents.

For now, these prompts are meant to help you come up with your own prompts, in relation to your dataset.

Get the picture? 

Use good communication skills and talk to a model as if it’s a very eager junior team member who knows everything about your business (if you’ve given it that information).

The Power of Human Communication

The irony of our AI age is that the way to win is to become more human. While AI might replace certain functions, success comes from enhancing our human qualities. 

The best copywriter I've worked with is a formerly Amish woman who gets incredible results with simple, three-to-four sentence prompts. Her success comes from her deeply human approach to communication.

Practical Tips for Working with Prompts

• Prompts go out of style as models are fine-tuned and filtered. What worked two months ago might not work today.

• Don't focus on exact wording - focus on intent and process.

• Treat the AI like a smart, eager colleague rather than a subordinate.

• Bad communication equals bad prompts. Good human communication skills translate directly to effective prompting.

• You can express any question or statement multiple ways - adjust your approach based on the responses you get.

Looking Forward

The process of communication will remain crucial even as the technical aspects of prompting evolve. 

The way we interact with AI is changing - many interfaces now allow voice input instead of typing. 

What matters isn't mastering specific prompt engineering techniques but developing strong communication skills.

The future belongs to those who can think clearly and communicate effectively. Ironically, English majors and others trained in language and thinking might become the rulers of the next era. The key isn't technical expertise but human understanding and expression.

Remember: Nothing about AI is inevitable. The narrative about AI as an existential threat is a red herring. The real consideration isn't the technology itself but how we choose to use it.

We have the opportunity to shape AI's role in our world, using it in ways that enhance rather than diminish our humanity.

It all starts with the language you use, in the form of prompts.

In the Age of AI, when the means of creation are available to anyone…

The perspective that reframes reality wins.

If you want attention and influence, you must cultivate a distinct point of view.

I wrote an essay on this here:

Your Perspective is a Power Law

If you thought the internet was noisy already, the volume is being cranked up past 11.

Open up any feed, from social media to news, and the firehose has become a constant, electrifying, pulsating, all-consuming, immersive strobe light beaming electrons into your eyes and skull at all times.

If you’re making, building, or selling anything online—how do you get noticed?

When anyone can perform magic tricks with the click of a button, why should anyone care about your stuff?

How do you get people to pay attention to you?

And how do you turn that attention into influence?

This is how:

Your Perspective is a Power Law

Talk soon,
Sam Woods
The Editor