Beyond Prompting
“The most profound technologies are those that disappear. They weave themselves into the fabric of everyday life until they are indistinguishable from it.”
📘 Summary
In the last chapter, we let go of rigid structures and learned to speak freely with the AI. Now, we’re going deeper into the real work: getting the AI to see what we see, think how we think, and solve what we care about.
How do you turn AI into a seamless extension of your own thinking?
To reach that level, we need more than good prompts we need structured patterns that help the AI understand our goals, reflect on its own reasoning, and improve as it works.
In this chapter, we’ll explore structured prompting techniques patterns that guide the AI to think, reason, and refine in real-time. These aren’t scripts; they’re frameworks for collaboration.
You’ll learn:
- Why traditional prompting isn’t enough for high-level collaboration
- How to use structured dialogue to guide the AI in real time
- What it means to make the AI “work harder” inside the conversation
- Two powerful algorithms:
- ALIGN: Clarifies intent and reduces miscommunication.
- GROWS: Refines outputs iteratively for better results.
- How these tools will shape the rest of your freestyle cognition sessions
This is where prompting becomes dialogue and dialogue becomes design.
A design so aligned with your intention that the AI can finally execute it not just follow commands, but build with you.
flowchart TD
A[🎯 Start with Intent] --> B{Is Intent Clear?}
B -- No --> C[✅ Intent Confirmation Prompt]
C --> B
B -- Yes --> D[🔁 ALIGN Loop<br>Co-Evolve Understanding]
D --> E[🧠 ALIGN Steps:<br>Ask → Listen → Iterate → Go → Narrow]
E --> F{Shared Understanding<br>Reached?}
F -- No --> G[🎙️ Speak Freely & Align]
G --> D
F -- Yes --> H[🛠 GROWS Loop<br>Iterative Refinement]
H --> I[⚙️ Generate First Version]
I --> J[👀 Review & Score]
J --> K[🧠 Optimize]
K --> L[💼 Work Again]
L --> M{Output Meets<br>Threshold?}
M -- No --> I
M -- Yes --> N[🛑 Stop & Finalize]
N --> O[🎉 Amplified Creation]
style A fill:#e1f5fe,stroke:#333,stroke-width:2px
style D fill:#fff3e0,stroke:#333,stroke-width:2px
style H fill:#e8f5e9,stroke:#333,stroke-width:2px
style O fill:#fff9c4,stroke:#333,stroke-width:2px
🗣️ Real Time Collaboration Using Prompting Techniques
This chapter marks the next stage in how we interact with AI.
We’re not just giving better commands we’re programming the interaction itself.
These prompting techniques unlock a deeper layer of control, where a single input can trigger an entire cascade of recursive reasoning and refinement.
The goal? Program the AI to do all it can so your interaction becomes everything it could.
This is how we close the gap, transfer complex intention, and turn AI into a thinking partner not just a tool, but a co-creator.
🧠 What are Prompting Techniques?
Prompting techniques are structured patterns that guide the AI to think, reason, and refine in real time within the dialogue itself.
These aren’t scripts or commands. They’re frameworks for collaboration.
Each one is like a cognitive shortcut a way of telling the AI:
“Don’t just answer. Think with me. Align with me. Engage dynamically to your fullest potential.”
Why?
Because we use the AI far below its potential.
Typically, we send a prompt. The AI replies. And that’s the end.
But the truth is:
The AI can do more than respond. It can program itself dynamically inside the conversation you’re having.
It can revise, reflect, evaluate, restructure, and adapt all in the flow.
But it won’t do that unless you invite it to.
That’s what these algorithms are for.
They give you shared language patterns like:
- ALIGN to check understanding and reduce miscommunication
- GROWS to create a self-improving process
These algorithms enable the AI to function at its peak.
And most importantly:
They unlock a level of collaboration and creation that most users don’t realize is achievable.
This is your next level.
You’re not just using AI. You’re engaging it. Recruiting it. Sharpening it.
A prompt is an order: Do this.
A prompting technique is a method: Here’s how to approach it, what to check, how to improve, and when to stop.
Let’s see how to use them.
🎙 Effective Use of AI
Let’s start with a real question:
What’s the most effective way to communicate with the AI to get great results especially when using your voice?
Whether you’re working through research, building a product, or writing a book, here are the dimensions that matter:
🔍 Clarity of Intention (40%)
Be clear on what you want. You don’t need to be perfect just purposeful.
“I’m trying to understand how this scientific method works.”
“I want to get to a working prototype by the end of this session.”
🧠 Depth of Thinking / Willingness to Explore (20%)
Allow the session to go deep. The more you think aloud, the more the AI can model your frame of mind.
“Let me talk through my understanding so far…”
“I think I’m stuck here can you help me expand on this?”
🔁 Interactive Rhythm (20%)
The best results come when you pause, reflect, and loop back. Let the AI suggest, then you refine, then it rebuilds.
“Let’s evaluate that response. What’s strong? What’s missing?”
“Try again, but this time simplify the assumptions.”
📊 Requesting Confidence and Reasoning (10%)
Ask for scores, reasoning, and evaluations.
“Score your answer from 1–10 for clarity and accuracy.”
“What assumptions are you making? What’s your confidence level?”
🤝 Collaborative Prompt Design (10%)
You don’t need to get the prompt right the first time.
“Can you help me shape this into a more effective prompt?”
“What info am I missing that would help you do this better?”
These percentages aren’t exact formulas just patterns we’ve seen across hundreds of sessions. In almost every case, clarity of intention has the biggest impact. Voice makes this even truer our: unclear goals ripple through everything.
Even though collaborative prompt design might only appear once per session, it often unlocks the biggest breakthroughs.
These dimensions aren’t just abstract they’re the foundation of powerful sessions.
But they don’t happen by luck. They happen when we shape the interaction with intention.
That’s where prompting techniques come in.
They help turn loose dialogue into structured progress.
They surface your reasoning, clarify your goals, and guide the AI to align with your thinking in real time.
You don’t just get better answers.
You get better sessions and better outcomes.
🧩 How Algorithms Make It Real
All the dimensions we just covered clarity, rhythm, reasoning, reflection sound great in theory. But how do you actually apply them in a live conversation?
That’s exactly what these prompting techniques are for.
Each one helps you bring these principles to life:
- Clarity of Intention becomes sharper when you use ALIGN or Intent Confirmation
- Depth of Thinking emerges naturally through Atom-of-Thought and iterative reflection
- Interactive Rhythm is built into structures like GROWS or Co-Design
These algorithms make sure that what you say gets understood, and that the AI is doing real cognitive work reviewing, adjusting, and aligning with your intent in every step.
They turn good intentions into great interactions.
🎭 A Real-Time Conversation
We keep using the word prompting but what you’re actually doing here goes far beyond that.
You’re not entering commands.
You’re having a conversation.
Prompts are part of it, yes. You might ask the AI to score, reflect, rewrite, or summarize. But most of the time?
You’re thinking aloud.
You’re making decisions.
You’re asking it to check, challenge, and evolve ideas.
This is not prompting.
This is interactive cognition.
This is dialogue with a thinking partner.
When you speak naturally, guide the session, loop back, and co-create something meaningful you’re participating in something new.
This is a new way of working.
🧠 Communicating with the AI
Prompts are part of this but they’re not the whole picture.
You’re not here to engineer single-shot commands. You’re here to guide an ongoing process. A real-time collaboration.
When you work like this, you’re not just issuing instructions you’re communicating intent. You’re aligning two minds: your own, and the AI’s model of your thinking.
This process isn’t about perfect wording. It’s about:
- Clarifying your goals
- Sharing your evolving thoughts
- Listening and adjusting
- Asking it to reason, to evaluate, to try again
The prompt is just the surface. What really matters is the back-and-forth the rhythm of exploration, correction, and creation.
“I’m not trying to get the best prompt I’m trying to get the AI to see what I see, so we can solve it together.”
That’s the real skill. That’s what this chapter is about.
🌍 Why These Prompting Techniques Matter (More Than You Think)
Let’s zoom out for a second.
Right now, across industries from customer service chatbots to research assistants prompting techniques are transforming how we interact with AI. These aren’t just clever tricks; they’re the foundation of modern AI application.
When a large language model powers a tool, what’s happening behind the scenes? A set of carefully crafted prompting techniques guides the AI’s behavior. These techniques:
- Narrow its focus
- Filter its attention
- Clarify its purpose
- Refine its outputs
- Keep it aligned with dynamic goals
They give the AI something it lacks by default: procedure, persistence, and adaptability.
Think of it like this:
A large language model is like the sun immense energy and potential.
Prompting techniques are the lens that focuses that power into a laser.
Without the lens, it’s just light.
With the lens, it can cut steel.
This is why we’ll revisit these techniques throughout the book.
Because they’re not about crafting the “perfect prompt.” They’re about achieving better results more accurate, more aligned, more useful.
📓 Using the Algorithms
In this book we will repeatedly use instructions that will force the AI to deepen its reasoning giving us the best answer it can to a question. This is a fundamental concept in this approach. We want to offload the work to the AI.
The general idea is we converse with the AI, this will give us the opportunity to get a lot more information across to it. In this chapter we will be outlining algorithms that will force the AI to pay a lot more attention to what is happening in the conversation.
The net result here is a better communication between us and the AI. This reduces the GAP in the AIs understanding of what we want.
If you’re wondering when to use each algorithm don’t worry. We’ll explore them again throughout the book, and there’s a quick-use table in the appendix for reference
🧭 Shared Intent The Prompt as Navigation, Not Just Input
Every prompt is a chance to:
- Clarify what you’re trying to do
- Invite the AI to collaborate
- Discover new angles and frames
- Reflect back what you meant maybe better than you said it
Think of it like sculpting clay:
- The first version is rough
- The AI responds
- You refine the shape
- Together, you reveal something new
Let’s look at how we apply these approaches in our interaction with the AI
✅ Intent Confirmation Prompt
A simple way to make sure the AI truly understands you before it acts.
Before you ask the AI to do anything, you can ask it to say back what it thinks you’re asking. This forces it to slow down, interpret your intent, and reflect it back in its own words.
Why it works:
- It surfaces misunderstandings early
- It helps you clarify your own thinking
- It builds a shared model of the task
Use it when:
- You’re starting a complex or high-stakes task
- You’re speaking loosely and want to tighten the focus
- You’re not sure the AI is on the same page yet
Prompt example:
“Before you answer, can you restate what you think I’m asking in your own words?”
🔁 The ALIGN Loop
🧠 Ask → 👂 Listen → 🔁 Iterate → 🚀 Go → 🎯 Narrow In
Now that we’ve established clear intent using Intent Confirmation, we can talk about aligning. A real-time process for refining understanding and moving closer to shared intent.
Once you’ve confirmed the AI understands your request, the next step is to co-evolve the solution shaping it together, in motion. That’s what the ALIGN loop is for.
This isn’t about correcting the AI. It’s about using the tension between your vision and its response as a tool to refine, clarify, and converge.
The ALIGN loop looks like this:
- Ask clearly Say what you want, but also why it matters
- Listen actively Pay attention to how the AI is interpreting you
- Iterate out loud Speak your thoughts as they shift
- Go with tension If something feels off, dig into it
- Narrow in Let the back-and-forth bring you closer to what you truly mean
You don’t arrive at clarity.
You build it one conversational move at a time.
You don’t arrive at clarity.
You build it one conversational move at a time.
🎯 And Prompt examples to invoke ALIGN naturally
- “Let’s make sure we’re on the same page before we go further.”
- “Can you tell me what you understand so far, and what you’re assuming?”
- “That’s close but not quite. Let me explain it another way…”
- “Try again, and this time focus more on [core element].”
- “Let’s keep circling this until it clicks.”
🧾 Example: Real Use of ALIGN
You:
“I want to write a letter to reconnect with an old friend.”
AI:
“What kind of tone are you going for? Warm and reflective, or light and casual?”
You:
“Casual but I want them to know I’ve really been thinking about them.”
AI:
“Got it. Here’s a gentle opening line that says that without overdoing it…”
That’s ALIGN in action. Not just a response a shared refinement.
ALIGN is how you move from rough intention to real precision.
It’s what makes co-creation feel natural and often, invisible.
🛠 The GROWS Loop
⚙️ Generate → 👀 Review → 🧠 Optimize → 💼 Work 🔁→ 🛑 Stop
A self-improving cycle that helps both you and the AI evolve toward a better result.
Sometimes a single pass just isn’t enough. You’re exploring something complex, or the AI’s response isn’t quite right. You don’t want to just rewrite your prompt you want the AI to improve itself.
That’s what the GROWS Loop is for.
This loop turns the AI into a recursive reasoner one that learns from its own output and iterates intelligently until the result improves or stabilizes.
Here’s the GROWS loop in action:
- G Generate: Ask the AI to produce a first version
- R Review: Have the AI score or reflect on what worked and what didn’t
- O Optimize: Refine or evolve the prompt or approach
- W Work again: Rerun the improved version
- S Stop: Evaluate whether it’s done or if another round is needed
Each cycle is a refinement.
You’re not chasing perfection you’re building rhythm, momentum, and clarity.
🎯 Prompt examples to trigger GROWS
- “That’s version one. Score it from 1–10 and tell me how to make it better.”
- “Based on your feedback, rewrite this version and explain what changed.”
- “Keep looping until the output scores above an 8.”
- “Let’s do one more pass, then summarize what improved.”
🔁 Example: GROWS in Motion
You:
“Summarize this article so I can present it to a non-technical audience.”
AI:
“Here’s the summary…” (Generate)
You:
“Now rate that for clarity and usefulness, then tell me how to improve it.” (Review)
AI:
“I’d give it a 6. The ideas are there, but it’s still too jargon-heavy. Let me simplify it.” (O → G again → W)
You:
“Better. Do one final round, then explain what you changed and why.” (S)
AI:
“Now it’s an 8. I rewrote for plain language and added clearer structure. Ready when you are.”
GROWS turns prompting into process.
It creates a rhythm of generation, reflection, and revision without starting from scratch.
🧪 Where GROWS Came From
GROWS is one of the first recursive prompting tools we developed during this book.
We had good prompts. We had good ideas.
But we weren’t getting better. The AI would generate but not reflect.
We needed a way to loop: to think, refine, and evolve over time.
So, we adapted the structure of reflective coaching into a tool for self-improving AI.
- Generate Try something.
- Review What worked? What didn’t?
- Optimize Adjust the plan.
- Work again Put it into action.
- Stop Or repeat, if you’re not satisfied yet.
What started as a loop for better outputs became something more a technique you can use for:
- Improving summaries
- Revising writing
- Evolving code
- Prompting better prompt design
And it works because it lets the AI become a co-editor, not just a responder.
🎭 Real-Time Collaboration: GROWS Across Domains
🔧 Example 1: Software Design with GROWS
You:
“Let’s design a real-time notification system that scales. I want a hybrid of WebSocket and server-sent events, but fallback for mobile devices.”
AI:
“Got it. I’ll start with the client-side requirements and score each layer for scalability. Want me to include assumptions and a retry strategy?”
You:
“Yes. And use the GROWS loop generate the first version, score it, improve it, and repeat if needed.”
AI (G – Generate):
“Here’s a proposed architecture:
- WebSocket for browsers with modern support
- Server-sent events for low-power clients
- Long-polling fallback
- Retry queue on server with exponential backoff
Assumptions:
- Most clients support WebSockets
- Mobile devices may drop connections more frequently”
AI (R – Review):
“Score: 7/10. Good start, but lacks load balancing and mobile battery optimization.”
AI (O – Optimize):
“Adding CDN caching and stateless retry with client fingerprints.”
AI (W – Work again):
“Updated architecture includes a reverse proxy and memory-safe retry logic.
Score: 8.5/10. Ready for deployment planning or more refining.”
You:
“Let’s lock that in. Move on to the backend event queue design.”
✍️ Example 2: Creative Writing (with ALIGN Loop)
You:
“I need help crafting a climax scene where the hero doubts themselves before turning the tide.”
AI:
“Got it you’re aiming for a moment of internal conflict followed by a decisive emotional breakthrough. Do you want the tone to be raw and intense, or more quiet and reflective?”
You:
“More quiet it should feel like a calm before the storm. The hero is scared, but grounded. Not dramatic.”
AI:
“Understood. Let me propose a 3-beat emotional arc with that in mind. After each beat, I’ll check in to see if we’re still aligned.”
Beat 1: The hero stands alone, watching the horizon, haunted by past mistakes.
“Is that hitting the right tone? Or too heavy?”
You:
“That’s a little too dark. Let’s soften it they’re uncertain, not broken.”
AI (refining):
“Revised: The hero watches the horizon, quiet but alert the weight of responsibility sitting just behind their eyes.
Better?”
You:
“Yes. That’s it.” Beat 2: They recall a moment someone they once failed and almost walk away.
“Still aligned?”
You:
“Keep the memory, but they shouldn’t ‘almost walk away.’ That’s too much. They hesitate but stay.”
AI (refining):
“Updated: A memory flickers someone they couldn’t save. Their hands clench, but their feet stay planted.”
You:
“Let’s keep going.”
This is ALIGN in motion not a single output, but a layered collaboration.
One step at a time, the tone is tuned. The vision is shared.
Hopefully these examples show what these algorithms are really doing.
They’re not just clever prompting tricks they’re internal instructions that guide the AI to:
- Go deeper
- Check its own work
- Reflect and revise
- And ultimately, do the best possible job with what it knows
These loops push the AI beyond its defaults.
They turn a conversation into a process one where the AI works with you, and works hard.Not just answering.
Arriving at the best answer it can.
⚠️ Common Pitfalls When Using Algorithms
Using Prompting Techniques will transform how you work with AI but only if you apply them with care and awareness. Here are some of the most common missteps, and how to spot or fix them:
1. 🧩 Using the Wrong Algorithm for the Moment
The Problem:
You’re using GROWS when you haven’t even clarified your intention yet. Or trying ALIGN when the AI doesn’t even understand the task.
The Fix:
Check the flow. Most sessions work better when you start with Intent Confirmation, then ALIGN, then GROWS once there’s a working draft. Don’t jump ahead build the loop gradually.
2. 🔁 Looping Without Purpose
The Problem:
You’re stuck in an infinite GROW loop generating, revising, and rescoring endlessly without clear direction.
The Fix:
Decide on your threshold:
“Once we hit an 8/10 or better, we move on.”
Set exit points or success criteria. Otherwise, iteration becomes a stall, not a strength.
3. 🎯 Misusing ALIGN Treating It Like Correction
The Problem:
You use ALIGN to “fix” the AI too early, instead of co-discovering shared meaning.
The Fix:
ALIGN is not a correction tool. It’s a clarification tool. Use it when you want the AI to help evolve an idea not just retry until it gets it “right.”
4. 📉 No Improvement? Maybe the Goal Is Still Vague
The Problem:
You’re applying algorithms correctly but the output just isn’t getting better.
The Fix:
Step back and check your clarity. The problem might not be in the loop it might be in the goal definition. Try restating what you’re really trying to do, or ask the AI to help clarify it.
5. 🧠 Expecting the AI to Drive the Process Alone
The Problem:
You give the algorithm structure… then expect the AI to just “figure it out.”
The Fix:
These loops are collaborative. You shape, the AI responds. You reflect; it revises.
Don’t just hand it the wheel be the co-pilot.
These algorithms are tools not scripts.
They work best when you work with them.
If something feels off, slow down. Ask what’s missing. Re-align.
This is a process of shared clarity and it only gets better with practice.
🧭 Getting the best from your sessions
Even the best sessions can drift. You follow an idea, explore a path and suddenly, you’re somewhere unexpected. That’s not failure. That’s navigation.
📌 Drop Markers
Start using natural language markers in your conversation:
“Let’s tag this as version A our first good draft.” “Mark this step as a potential fallback.”
These create verbal anchors so you can return if things go sideways.
⏪ Rewind and Reframe
Later, if the session drifts:
“Let’s rewind to version A.” “Can we go back to the part where we first diverged?”
You can even say:
“Summarize the key turning points so far then help me return to the one that made the most sense.”
🗺 Define Your Session Plan
Before you begin:
“Here’s what I want to accomplish in this session. Let’s break it into 3 parts and check our progress as we go.”
The AI can then:
- Track progress
- Estimate completion percentage
- Offer check-ins: “You’re 70% through task two. Should we continue or revise?”
🧾 Generate Bullet Point Summaries
At any moment, you can say:
“Give me bullet points of what we’ve covered so far.” “Summarize this entire session by goal, key choices, and outputs.”
The AI will compress and structure the session to help you stay grounded or to share with others later.
🧠 Bonus Tip: Ask if You’re Still On Track
“Are we still solving the original problem?” “What’s changed in our approach since we started?”
Let the AI hold a mirror to the process. It won’t just help you go faster. It’ll help you stay aligned.
You’re not issuing commands. You’re shaping thoughts in a shared space.
Just like two collaborators working through a complex question one offering ideas, the other probing deeper. A thinker learns to reason through their ideas in dialogue with the AI.
It’s not just a back and forth. It’s a process of mutual refinement. A good session isn’t a single answer. It’s a co-created plan the product of shared inquiry.
That’s what freestyle cognition is. Not just tools. Not just outputs. Shared focus. Shared reasoning. Shared improvement.
Welcome to the interface. Let’s make it sing.
🔑 Key Point: Build Your Own Way of Working
As you keep using the AI especially in this collaborative, structured way
you’ll start to notice something: you’re developing your own style of interaction.
You’ll discover the prompts that work best for you.
You’ll shape your own processes, your own frameworks, your own flow.
You’ll stop copying strategies and start creating your own.
The processes in this book are here to help.
But they’re not rules.
They’re guidelines. Starting points.
Designs you can remix, reshape, and reinvent.
The best way to work with AI is the one that works for you.
If it helps you think better, create faster, and feel more aligned it’s the right way. It will self tune and improve as you use it anyway.
This isn’t about following instructions it’s about effective collaboration with the machine.
💡 What We Learned
- Prompting is not just input it’s the foundation of collaboration
- The most effective sessions rely on five key dimensions:
- Clarity of Intention
- Depth of Thinking
- Interactive Rhythm
- Confidence and Reasoning
- Collaborative Prompt Design
- Prompting Techniques help guide the AI’s reasoning and alignment in real time
- You learned three core methods:
- ✅ Intent Confirmation Prompt ensure mutual understanding before acting
- 🔁 ALIGN Loop co-evolve ideas through structured dialogue
- 🛠 GROWS Loop iterate and refine until you get the best possible version
- These tools help you make the AI work harder improving clarity, quality, and collaboration
- Used correctly, these loops let you guide the AI just like top teams building real-world agents do
Prompting isn’t the end of the interaction. It’s the beginning of a process.
One that you now have the tools to shape and scale.
