Welcome to Day 4...

When you watch today's stream, you’ll notice a major shift in how I am approaching these sessions.

I am officially retiring myself as a "teacher."

From here on out, think of AWT Backstage less like a classroom...

And much more like an artist’s studio... an "AI Dojo" of sorts.

Years ago, when I practiced Aikido, the best learning didn't come from a textbook.

Someone would step onto the mat, model a movement, and then everyone practiced together.

There was an element of structured play to the whole thing.

It was like learning how to play a new game as you built the skills to play it.

That is exactly the environment we are building here.

I am going to model the art of using Large Logical Machines, and I want you in the game.

Here is the Casual Magic under the hood in Session 4:

• Banishing the Ghost of the Developer: We introduced B.E.N.T.O. yesterday, but as you'll see in today's assets, we are already throwing the acronym away. Memorizing acronyms feels like homework. Today, we shift from a "manual transmission" to an "automatic" using the Conversational Canvas.
• The Portable Skeptic: This is your silent, backstage bodyguard. It stress-tests your logic (autonomously asking the "impossible" questions) before the machine even starts working. This is how we completely eliminate "prompt engineering."
• Validation from the Source: I spent time this week in calls with Google and DeepMind developers. I asked the hard questions about data privacy and edge-node processing. The answer? The data sovereignty model we are building works. It is entirely possible to have an AI ecosystem that doesn't care about scraping your private data. 🤓

My Ask of You (The Workbench)

I have an absolute arsenal of custom Gemini Gems and logic tools running my backend (we are currently on Day 49 of this OS running my business).

I am more than happy to package these up and share them with you...

But I need to know you are in the game.

If you are watching these videos, drop a comment.

Hit a like button.

Post your messy experiments in the Workbench.

Even if you are sitting in the back going, "I have no idea what is happening, but I'm here," let me know. 😅

I will build tools for active builders, but I need you to signal that you are ready to play.

The future is coming fast (we are targeting Summer 2026 for free personal Nitwyt™ nodes).

Let's get building...

AI First... Always Human.

If you watched the Session 3 live stream (attached here)...

You caught a classic human moment.

I spent a significant chunk of the video passionately explaining a core concept...

Completely on mute. 😅

Instead of hiding that or trying to edit it perfectly away, we are going to use it.

This is exactly why we rely on our digital tools.

As humans, we are beautifully messy.

We get tired, we make mistakes, and we forget to unmute the microphone.

Our goal is not to become perfect machines.

Our goal is to set up a reliable environment where our AI (acting as our "Silent Partner") handles the heavy lifting, allowing us to just be human.

We call this...

AI First... Always Human.

The Visual Shift of Casual Magic

During that silent stretch of the video, I was showing you a series of slides.

If you watched closely, the entire mood, color, and art direction of the presentation shifted on its own.

When we were discussing the "Chatbot Delusion" the visuals were chaotic and unappealing.

(One looked exactly like taking a five-star meal and dropping it all onto one plate.)

But the moment the message shifted to the Bento concept the visuals instantly snapped into a calm, warm, and pristine layout.

Here's the magic...

I didn't tell the machine (AI) to change those specific colors or lighting.

By simply arranging our thoughts into clear boundaries, the AI ("Silent Partner") applied basic common sense.

It understood we were moving from the stress of a blank screen to the absolute relief of perfect organization, and it visually built that relief for us.

Your First Recipe Card

We want you to experience that exact same sigh of relief today.

Here is your very first Bento Recipe Card.

Think of this as your training wheels.

You do not need to write a complicated paragraph or learn a new technical language.

Simply copy the recipe below, and drop your daily riff into the [NEEDS] compartment.

Watch how your AI snaps your chaotic morning into perfect structural order.

(Note: If you are planning your afternoon instead of your morning, simply change the context in the [ENVIRONMENT] compartment).

YOUR BENTO RECIPE CARD:

[BOUNDARIES] You are my Silent Partner. Use zero technical jargon. Keep the tone calm, direct, and highly supportive. Do not ask me follow-up questions; simply execute the task. [/BOUNDARIES]

[ENVIRONMENT] It is morning. I am feeling overwhelmed with my schedule. I have exactly 15 minutes to organize my day before my first commitment, and I need immediate clarity. [/ENVIRONMENT]

[NEEDS] {{Drop your Fuel here. Paste your riffs, thoughts, emails, or anything else that represents the chaos of your upcoming day directly into this space}} [/NEEDS]

[TASKS] Read my Fuel. Snap this messy data into clear, structural order. Categorize my items into three simple buckets: what I must execute immediately, what I can delegate or push off, and what I can safely ignore. [/TASKS]

[OUTPUT] Give me the final result as a clean, highly readable, bulleted checklist. [/OUTPUT]

Try it out today.

Instead of thinking of the chat window as something you have to converse with, try thinking about it as your own personal Canvas.

How you structure your logic is the art. 🤯

You'll see even more tomorrow.

Howdy!

I just wrapped up an unlisted live stream rehearsing Day 2 of the new challenge, and I’m dropping the raw, quick-and-dirty video here for you first.

In this one, I’m pulling back the curtain on how I'm actually building this content.

You'll see me actively using NotebookLM's podcast studio...

I'll show you how I'm using its critique features to improve the lessons...

And I’m sharing the exact behind-the-scenes logic structures (the LEAP files) I'm using to guide the AI.

As you watch me just acting as a content producer building out this challenge, I'm hoping you catch the underlying message of Day 2...

Logic...****Not in the way you think.

When people hear the word "logic," they immediately think of complex math or glowing green computer code.

(But logic is actually just common sense.)

When you organize your kitchen so the coffee pot is right next to the mugs, you aren't "programming" your house.

You are simply using spatial logic (i.e., "common sense") to make your morning easier!

Right now, the biggest source of digital stress is staring at a blinking cursor, trying to figure out how to "prompt" an AI.

We need to drop the idea of "prompting" entirely.

Instead of typing a messy paragraph guessing what the machine wants, we just need to load a structure (like a simple "Recipe Card" gives structure).

When you give a Large Logical Machine a clean, common-sense boundary (The Goal, The Tone, The Facts)...

It acts as your silent partner and does the heavy lifting instantly.

You stop guessing.

(And more importantly, you get your afternoon back.)

Check out the backstage video below to see the slide decks, the custom NotebookLM prompts, and how I'm using this exact "canvas" logic to build the challenge itself.

Let me know what you think of the direction...

Because tomorrow, for Day 3, we are going to take this common-sense canvas and show you a very specific way to pack it.

(Hint: Get ready for the Bento Box). 🍱

Howdy!

I’m currently in the middle of building out the new 7-Day Challenge, but I wanted to drop a quick-and-dirty 20-minute backstage video here for you guys first...

It includes some slides and a Notebook LM presentation I’ve been playing with to map out the "Day 1" concepts.

Here is the core reason why I'm building this...

Right now, almost the entire world is trapped in the "Chatbot Delusion."

We open up an AI...

We treat it like an incredibly smart intern...

We ask it to write an email...

Maybe to summarize a document...

Or help us build a basic automation, vibe-code, etc.

There's nothing wrong with that.

(It's the natural first step.) 🙌

But it's a fraction of what is actually sitting in front of you.

We think we are talking to a Large Language Model.

(We aren't.)

We are operating a Large Logical Machine. 🤯

When you just talk to AI off the cuff, the machine simply mirrors your own unstructured logic back at you.

When it gives a bad output... That isn't a bug in the AI.

(It's a feature.)

It's perfectly reflecting the gaps in the logic it was handed.

But when you realize this... everything changes.

If the AI already has access to all the knowledge in the world...

Then the only barrier left is common sense and your own logical creativity.

You get to stop worrying about what you know...

(Instead, you start cross-referencing patterns to build exactly what you need to get your time back.)

You can weave together radically different ideas to build things that simply do not exist yet...

Because no one else has logically thought to connect them!

We don't need to learn how to write "better prompts."

We need to drop the idea of "prompting" entirely, and learn how to structure a better request.

Check out the video below to see some of the imperfect, raw patterns I'm playing with to help people make this shift.

Let me know what you think of the direction...

Earlier today I did a YouTube live stream...

(FYI... When you listen to this stream turn your volume all the way up. I messed up the audio levels on my end, so the sound is incredibly low.)

To be honest, I recorded this primarily to use as raw data for our AI training anyway, but the visual demonstration inside this video is too important to hold back.

If you have ever felt the frustration of getting your AI moving in the right direction, only to watch it completely "drift" and forget your original rules two prompts later... you need to watch this.

In the video, I do a live, side-by-side comparison of normal "prompting" versus a completely different method.

It is a technique called Stereo Containerization.

Instead of just typing a paragraph and hoping the AI understands what you want...

You are going to see me put the AI's "Rules" in one specific digital box, and the "Raw Data" in a completely separate box.

When you structure it this way, the AI cannot drift... It just executes.

Even with the low audio, you can watch exactly how this new style of prompting works on the screen.

I highly recommend practicing this technique today, because this is the exact foundational skill we are going to use over the next few weeks.

In fact, once I finish assembling the core engine of the Digital One Builder's Kit, one of the very first things I am going to do is feed this exact video into the kit and force the AI to automatically build out our course materials from it.

You can watch the recording here (remember to crank the volume!)

Two quick housekeeping notes:

1. For everyone in the Workbench: I will be dropping a lot more videos like this over the next few weeks as we build out this system. (And I promise I will test the microphone first next time 🤣).
2. For those who grabbed the Builder's Kit Already: Thank you for supporting the build! We are spinning up a dedicated, private space just for you. All the specific instructional videos, .leap files, and module templates for the kit will be dropped in that specific area so everything is organized in one place. Keep an eye out for that space to show up over the next 24 hours or so.

Watch the demo, test out the Stereo Containers on your own AI today, and drop any questions you have below!

I can't wait to deploy in my businesses.

I'm glad to be part of this project that Mercer is putting together, and I'm looking forward to enjoying the experience with everyone.

I learned about the LEAP format in January when Mercer showed it to me in Nashville and ever since that time I've been excited to deploy it to build a customer chatbot in my business - a hypnosis training company. I'm thrilled to share the early results.

THE FIRST STEPS:
Mercer basically told me to record a long riff / rant about my business, and what the function of the chat bot should be. I went for a walk, had my phone in my hand and just blabbed all about the business, what I wanted the bot to do, how I wanted to create an obviously useful experience for customers / visitors, and how it should help them by segmenting into one of two possible buckets as early as possible in the convo.

I got the transcript and hopped into a Google Meet with @mercer. He got me to bring one other asset, which was a lame Google Doc that I created a while ago to be a basic bot training doc. We had me do a 'mini riff' to explain to the Gem that I'm attaching two files, one is the long riff, and one is a doc I built previously.

The output was INCREDIBLE. The gem basically gave me 3 LEAP files that we immediately tested in NotebookLM and Gemini.

The tone, the intention, the helpfulness ... the test responses were really amazing. But it still lacked knowledge.

So I asked Mercer if I should record myself updating the LEAP docs once I was ready.

And I did just that yesterday!

I just went for another walk, and recorded a riff all about the main product, the Mike Mandel Hypnosis Academy.

I then asked in my original "RIFF TO LEAP" chat, "Hey, here's some more info about our main membership that might be useful".

It produced a bunch more output (two new LEAPs and updates system instructions).

I am finally understanding how to build this stuff, and it's actually fun because all I have to do is go on a wild rant about things, record it all, and feed it to the Gem.

I am attaching the full video showing the exact process I used to update my initial LEAP documents so that we can all benefit from any feedback to me here in public, so everyone learns faster.

Hey everyone!

​I’m here because of a conversation with Chris Mercer that acted as a bit of a 'glitch in the matrix' moment. It made me realize that running a conversion optimization agency the 'traditional' way is a fast track to burnout and inefficiency.

​I’m not just looking to tweak a few workflows; I’m here to completely reinvent how my agency functions under the influence of AI. The goal? Automating the heavy lifting—specifically the copywriting and technical setup—while simultaneously dreaming up entirely new service offerings that weren't even possible twelve months ago. I want to move from 'manual labor' to 'algorithmic impact.'

​If you’re also in the process of tearing down your current business model to build something smarter (or if you just know how to make AI write copy that doesn't sound like a robot trying to sell me insurance), I’d love to connect.

​Excited to see what we can break and rebuild here!

Any ideas?
I have about 5 Gemini gems that I want to collaborate to help be more effective. I have come across one snag. The models I know of do not allow api access to a private gem.

GPT told me I could somehow host the Gem in Discord and then use n8n to connect it to Gemini. That is something I am going to try.

Currently, I am using the old scratch pad method. At the end of each conversation with a gem, it puts together the relevant information needed for the group. Then I go to the next gem and it reads the document and away we go.

It is much slower than it should be but it is better than just going back and forth myself.