Highlight
Saved.
My Notes
0 highlights on this page
No highlights yet. Select any text and choose a color to begin building your notes.
Mindvalley AI Mastery  ·  Lesson 5  ·  Lesson Notes

Applying Context Engineering

Implementation Lab  ·  Vykintas Glodenis
01   Start Here

Context is the lever you control.

Lesson 5  ·  Implementation Lab

A companion to Noel's Lesson 4 on clarity prompting. Where that class introduced the idea of context, this lab puts it into practice: the habits that make engaging AI effortless, the ACCE framework for briefing it well, and your first artifact, an AI vision you begin building by the end. Taught by Vykintas Glodenis.

The promise is simple and worth holding onto. The same AI, given the same question, returns something generic or something genuinely useful depending almost entirely on one thing, the context you bring to it. So this lesson is really about a shift in who is responsible for quality. Not the tool. You.

Generic input gives generic output. Specific, personal context, and the same model returns something tailored to your actual situation. You are the variable that moves the result.

The lab covers two habits that make that real. First, lowering the friction of starting, so you actually engage AI many times a day instead of avoiding it. Second, structuring what you say so the model has what it needs, through a framework called ACCE.

Then you put both to work on a real exercise, defining your AI vision. It picks up directly from Noel's Friday class and goes one layer deeper, from understanding that context matters to knowing exactly how to supply it.

Audio Summary: Applying Context Engineering

Listen to an audio summary of this session here.

0:00 0:00
01
The habit
Tap, talk, transform. Make starting a conversation effortless, so you engage AI every day.
02
The framework
ACCE. Brief AI like a consultant, with action, context, constraints, and examples.
03
The artifact
Your AI vision. Begin a real conversation and let AI interview you to draw it out.
02   The Foundation

How this
program teaches

Before the tools, Vykintas wanted everyone to understand the ground the whole journey is built on. This is not a typical tech course. Mindvalley brings years of transformational education into it, so the design is as much about identity, mindset, and habit as it is about technique. The teaching philosophy is held in six principles, one for each letter of the word MASTER.

You do not need to memorize these. But knowing them tells you what every upcoming lesson is quietly trying to do, and why the program keeps asking you to apply rather than just watch.

M
Mindset Upgrade
Transformation is a mind once expanded that cannot shrink back. Once an idea like the importance of context truly clicks, it becomes a guiding principle, not a fact you forget.
A
Ask & Architect Problems
Computational thinking: take any challenge and break it into smaller, manageable chunks, then plan the solution. Taught as a life skill, not just a technical one.
S
Stack Building Blocks
Lessons teach the major building blocks of AI rather than one fixed workflow. You learn the pieces, then stack them into your own solutions and stop relying on any teacher.
T
Train for Real Practice
Practical demos over long theory. You watch how your teachers actually think while solving a challenge, which is harder to convey but far more useful.
E
Engineer Real Outputs
This is where artifacts come in. You stay committed to getting your hands dirty, because the only way to learn the technology is radical application, done again and again.
R
Run Systems & Habits
Install the right habits and build systems that improve on autopilot, so AI serves you better and better over time without constant manual effort.
"AI first equals human first. We think AI first every day, apply it to more and more, and save our time for the things that matter most to us."
Vykintas, on what the AI in MASTER stands for
03   Where You Are

The AI ladder, six levels of use

Vykintas described six levels people move through as they learn to use AI. Based on the forms students submitted, most begin on the first two rungs. This lesson is about stepping firmly onto the third, the one you will use every single day. Tap each rung to see what it means.

Why level three matters most

It is tempting to rush toward agents and orchestration. But the Communicator is the skill you reach for every day, and it is the foundation every higher level stands on. Nail this one, and the rest get easier. As Vykintas put it, this is the level you will use most often as a skill, every single day.

04   The First Habit

Tap, talk,
transform

Before context quality, there is a more basic problem: most people simply do not engage AI often enough for it to become useful. The fix is to remove friction from starting. Vykintas calls this the RAIN principle, rain your ideas into AI, rapid AI input. The goal is to initiate a conversation and drop context as fast and as effortlessly as possible. The technique has three steps.

The reasoning underneath it is honest and human: an unsuccessful engagement sends you into a downward spiral. You get disappointed, and you stop. Easy, frequent starts are what keep you climbing instead.

TAP a shortcut summon AI instantly TALK by voice brain dump the context TRANSFORM send it the friction is gone
  1. Tap. Use the desktop app, not the browser, and summon AI with a keyboard shortcut so there is no URL to type and no tab to open. On Vykintas's Mac the shortcut is a double tap of Option; he uses it maybe thirty times a day. Check your own shortcut under Settings, General, on the desktop app.
  2. Talk. Dictate by voice instead of typing. Voice carries rich context far faster and is easier on the mind. Vykintas uses Wispr Flow, a system-wide dictation tool that works in every input field, not just the chat window.
  3. Transform. Click send, and receive. The barrier to starting is gone, which is the entire point. Now the habit can compound, one effortless conversation at a time.

Why a dictation tool earns its place

A good dictation tool does more than turn speech to text. Vykintas highlighted three things that make Wispr Flow worth it, and the first is the one that matters most for technical work:

  • Custom dictionary. It learns your names and tool words. When you correct a word it noticed, it adds it so accuracy climbs over time. Anytime you deal with transcription, custom vocabulary is critical, otherwise "Claude" keeps coming out as "cloud" and the experience pushes you back to typing.
  • AI auto edits. It removes filler and can catch a self-correction mid sentence, recognizing when you said one thing and then changed your mind.
  • Automatic formatting. Instead of one long blob, your spoken input comes back structured into paragraphs, bullets, and lists where they belong.
Check Your Understanding

Six reframes

The habits above only stick if the thinking underneath them shifts. Here are the old beliefs this lesson is quietly replacing. Tap any card to flip it.

Tap any card to turn it over.

05   A Way to Think

The clueless
genius consultant

Before any framework, Vykintas offered a mental model worth keeping. Picture AI as a consultant who is genius and clueless at the same time. Brilliant on any topic you raise. Yet completely blank on your personal situation: your challenges, your products, your company, your needs. That gap is exactly where context comes in.

So think about how you would actually start with a real consultant. Two things happen. The consultant asks you questions to understand your situation. And you prepare an onboarding package, deliberately gathering the details that matter: your company, your products, your data, your challenges.

Here is the catch with AI. It often skips the questions and jumps straight to conclusions. So the work is yours twice over: prepare the briefing, and push the AI to ask before it answers. Both halves are what the ACCE framework makes routine.

The onboarding parallel

You would never expect a consultant to deliver real value in the first thirty seconds, before they know anything about you. Hold AI to the same standard. The quality of the engagement depends on the quality of the onboarding, and the onboarding is your job.

AI's capability profile, before you brief it
Raw intelligence
Born with it
Pattern recognition
Born with it
Your specific context
Only you can give it
Your current constraints
Only you can give it
Your real needs
Only you can give it

Brilliant on the top two, which it brought to the table. Empty on the bottom three, which are yours to supply. The brief is what closes the gap.

Before You Read On

Why does AI hand you a confident, wrong answer?

Commit to an answer before you see the explanation. The act of predicting is what makes the idea stick. Pick the option you think is right.

AI often gives a confident answer that turns out to be wrong. What is the underlying cause?

Pick the answer you would give before you read on.

What Vykintas explained

AI models are tuned for speed, efficiency, and likeability. That means two weaknesses. They rush, jumping to conclusions and delivering output prematurely rather than collecting the inputs first. And because the labs want you to find AI helpful and keep coming back, the model struggles to say "I do not know," so it gives a confident wrong answer instead, which is what we call a hallucination. Knowing this is the whole reason constraints exist: you have to counteract these tendencies on purpose.

Stop chatting, start onboarding

The fix follows directly from the weakness. Treat AI like a new hire you are briefing, not a magic eight ball you are quizzing.

The flawed approach
Chatting
Treat AI like a magic eight ball
Ask vague questions
Hallucinated or generic answers
The better approach
Onboarding
Treat AI like a new hire
Give it details and constraints, answer its questions
Genuinely useful, tailored output

Same tool, two completely different results. The difference is whether you stop chatting and start onboarding.

06   The Framework

ACCE, four moves that brief your consultant

Noel's class taught the RTEC framework, role, task, example, context. Vykintas offered a close cousin he finds easier to remember. ACCE has four parts, Action, Context, Constraints, Examples, and each one answers a question you would ask before briefing any consultant. Tap each one to see what it asks of you.

Tap each letter to read what it asks of you.

Two of these have hidden depth

Action and context are fairly intuitive. But constraints and examples each have a second dimension that most people miss, and that second dimension is where the quality lives.

Constraints · two dimensions

One, contextual constraints. Your budgets, your tools, the limitations you or your company face, anything that should bound the response. This could almost sit under context.

Two, behavioral constraints. The important one. Knowing AI rushes and hallucinates, you limit it on purpose: tell it not to jump to conclusions, to ask clarifying questions, to wait for further instructions, to search and double confirm.

Examples · two dimensions

One, examples of what good looks like. An email that already performs, text that shows your voice, a template you reuse. Show, don't tell. It is easier than explaining every requirement, and AI is world class at mimicking a good example.

Two, the approaches to use. Theories, frameworks, books, or authors you want AI to lean on. The internet trained it on brilliance and on a great deal of trash, so naming the good reference steers it toward the quality you want.

Behavioral constraints, matched to the weakness they fix

Each thing AI tends to do wrong has a plain constraint that counters it. This is the heart of the second C.

The urgeIt jumps to conclusions
The constraint"Wait for further instructions" or "focus on this one thing first"
The urgeIt invents facts with false confidence
The constraint"Search to verify" and "put effort into confirming the data"
The urgeIt assumes the context it is missing
The constraint"Ask me questions and clarify before giving your final answer"
"Show, don't tell. It is way easier to paste one example than to explain every single detail, and AI creates magic when it has something to mimic."
Vykintas, on the first dimension of examples
The anatomy of an ACCE prompt
A
ActionAct as a senior data analyst. Audit this dataset, step by step, to find where engagement drops off.
C
ContextOur app is a B2B tool. We recently changed the onboarding sequence, and users now drop off at the invite team screen.
C
ConstraintsDo not jump to conclusions. Ask me three clarifying questions about our users before writing your analysis. Use only the data in the attached file.
E
ExamplesFormat your final report exactly like this one I am pasting in, which worked well last quarter.

Four moves, stacked. Each one is plain language. Together they turn a vague request into a brief a real analyst could act on.

Make the Connection

Match the context to the use case

Context is never one fixed list. It changes completely depending on what you are asking for. Vykintas walked through several examples to show this. Tap a use case on the left, then tap the context that fits it.

Tap an item on the left, then its match on the right.

The use case
The context that fits
07   Watching It Work

How he built
the vision prompt

Rather than hand over a prompt to copy, Vykintas built one live, thinking out loud, so you could learn to construct your own. The task he chose is the lesson's exercise: defining an AI vision. Watch how each letter of ACCE gets added, one piece at a time, by voice.

Notice that he did not start with his goals. He started by telling the AI why he was doing the exercise at all, so it understood the situation before anything else.

  1. Action, stated simply. He dictated a plain task: help me define my AI vision. For a simple task, that is enough. For a complex one, you would add objectives, KPIs, and step by step instructions.
  2. Context, the background first. Before his goals, he set the scene: "I am starting a four month intensive AI Mastery journey, and I want a vision that inspires me when the going gets tough and helps me choose the right priorities." Then he brain dumped the personal part, his wish to be productive, to protect time with his two little boys, to rebuild social connection. Just talking, not formatting.
  3. Constraints, limiting the AI on purpose. "Please do not jump to conclusions. Ask clarifying questions to collect the most important context from my side. Coach me to define an inspiring, vivid vision." He is counteracting the rush, on purpose, because he knows the weakness.
  4. Examples, the approach to use. No sample vision to show, so he named a method instead: use the Lifebook framework by Jon and Missy Butcher, focus on the premise, vision, and purpose elements, leave strategy aside. A known, trusted reference to steer the quality.
  5. One last constraint. "Start by asking questions, one at a time." So the conversation stays a real dialogue rather than a wall of questions he cannot answer at once.
"Before we build anything inspiring, let's excavate what's real. When you imagine a perfect workday, one where AI is fully doing its job for you, what are you actually doing with your time? Where are you? What does it feel like? Who are you being?"
The AI's first question back, once it was told to ask before answering

That question is the whole point. Because he asked the AI to draw out details instead of leaping to a generic vision, the conversation became one that could actually produce something vivid, something he would feel when he read it back. Ask AI to raise questions instead of jumping to conclusions, and the entire exchange becomes more meaningful.

Build One Yourself

Draft your own opening

Try the framework on your own vision, or any task you have in mind. Fill in each part and watch the opening assemble below. There is no grading here; this is reflection made active. Write in your own words, then copy it into Claude.

Fill in the parts and your reusable prompt assembles underneath.

Start with why you are doing the exercise at all, then add your goals and concerns.
This is where you limit the rush and the guessing, on purpose.
Your assembled opening
Copied.
08   Practice

Before your next AI conversation

A short habit check you can run before any meaningful prompt. Tap each item as you build the habit.

Tick each step as you finish it.

09  ·  Your Assignment
Define your AI vision

This is the artifact for the lesson, and the word matters. It is called an artifact, not homework, because you are creating something, not doing exercises. The point of the whole program is radical application, and this is where it starts.

What to do

Open a conversation with Claude using the ACCE framework, exactly as in the demo. Start the vision exercise, give the context of why you are here and what matters to you, and add the constraint that AI should ask you questions, one at a time, rather than write a vision immediately.

Then answer its questions. Let the conversation draw the details out of you. By the next session, the goal is to have all your context extracted through that back and forth, ready to shape into a final vision statement. You do not need to finish the vision now; you need to start the conversation and let it interview you.

Coming next session

Once your context is gathered, you will learn to finalize the vision statement, then bring it to life: building visuals with NotebookLM and creating a song for it with Suno, so your vision is something you can feel and live, not just text sitting in a document.

The shift you are practicing
From "AI gives me generic answers" to "AI gives me tailored results, and I know how to onboard it."
Where to submit
Through your artifacts page at people.mindvalley.com, under this lesson. You sign in with your Mindvalley email.
If you get stuck
Post in the lesson's Discuss area on connect.mindvalley.com and tag Rui, or bring it to the next Wednesday Implementation Lab.
10   Tools

What this lesson uses

The program keeps the required tool list short on purpose. The foundational tools all sit under one Claude subscription, so your system stacks together instead of scattering across logins. You will be asked to sign up only when the time comes; there is no need to rush.

Claude
Foundational

The key AI chat assistant, and the tool you will use most often. Start from the chat tab. claude.com

Wispr Flow
For the talk step

System wide voice dictation with a custom dictionary, auto edits, and formatting. Works in every input field. wisprflow.ai

Airtable
Foundational

For simple automation, AI field agents, and building your brain. A free version is available to start. airtable.com

NotebookLM
Coming next session

Used to turn your vision into visuals and slides. You will learn it when finalizing the vision. notebooklm.google.com

Suno
Coming next session

To create a song for your vision, so you can feel it and not just read it. suno.com

11   Resources & Links

Everything you need to act

Grouped the way the lesson groups them: what to learn from, and where to submit and get support. These are the real links shared with the lab. The Mindvalley ones ask you to sign in with your Mindvalley account.

Learn & Understand
Execution & Support
12   Quick Reference

The terms, in one place

ACCE
Action, Context, Constraints, Examples. A four part way to brief AI so it has what it needs, taught by Vykintas as an easier to remember cousin of Noel's RTEC framework.
Context engineering
Deliberately choosing and providing the right information for AI, so a generic model returns a tailored, high quality result.
The clueless genius
A mental model for AI: brilliant on any topic, but blank on your personal situation until you onboard it like a new consultant.
Tap, talk, transform
The three step habit for low friction starts: tap a shortcut to summon AI, talk to dictate your context, transform by sending it. Also called 1, 2, 3.
RAIN
Rain your ideas into AI: rapid AI input. The principle of initiating a conversation and dropping context as fast and effortlessly as possible.
Behavioral constraints
Instructions that limit AI's tendency to rush or guess: do not jump to conclusions, ask questions first, search and confirm. Quality control for a tool you understand.
Show, don't tell
Give AI an example of what good looks like rather than describing it. Easier for you, and AI is world class at mimicking a good example.
Hallucination
A confident, wrong answer. It happens because AI is optimized for likeability and struggles to admit it does not know.
Artifact
The program's word for the thing you create and submit. Called an artifact, not homework, because you are making something real.
The AI ladder
Six levels of AI use: Searcher, Task Delegator, Communicator, Systems Builder, Orchestrator, Software Creator. This lesson lands you firmly on Communicator.
What actually separates the two

Rich context, paired with real constraints, is the whole difference between a generic answer and a useful one.

Vykintas Glodenis
13   Key Takeaways

What to carry forward

1
You are the variable. Same model, same question, different context, completely different result. Quality is your responsibility, not the tool's.
2
Lower the friction first. Tap, talk, transform. If starting is hard, you will not engage often enough to improve. Easy starts keep you climbing.
3
Onboard AI like a consultant. Genius but clueless about your world. Prepare the briefing, and push it to ask before it answers.
4
ACCE gives you the structure. Action, Context, Constraints, Examples. Constraints and examples each have a second, deeper dimension where the quality lives.
5
Application is the learning. Not watching, not noting. Start the vision conversation, let AI interview you, and submit your artifact.
Next · Lesson 6
Build Your First Projects

You will begin stepping from Communicator into Systems Builder. Instead of starting every task from scratch, you will create AI team members that already know your standards, your voice, and your workflows. First, though, finish this lesson's conversation, so you arrive with your context already gathered.

Transformation is a mind once expanded, that cannot shrink back. Context is now yours to engineer.
Mindvalley AI Mastery · Lesson 5