Make Your AI Agent Think Critically: 3 Markdown Files + 1 Sub-Agent

Vicky Zhao

Founder| Ex-Google & Mgmt Consulting

Build 1 agentic system, 3 agentic workflows in 3 wks infused with your judgement

You already use LLMs and know AI agents like Claude Code are powerful.

But you're not sure where to start. What's the difference between chat and agent? What and how to build your agent? What context to give it? How to get it to use your expertise and judgment? And how to check its answer so you actually feel comfortable putting your name on the output.

This is the part that tutorials skip. Because this requires deep thinking and frameworks that systematically extract and leverage your expertise and experience. This is not about which button to press.

This cohort is where we do the deep thinking and build an agentic system that doesn't default to average.

In 3 weeks, you will build 3 real Claude Code workflows while learning how to choose the right use case, extract your judgment, verify the output, and turn what works into something automated so you can focus on strategic work that matters.

The goal is not to become technical. It's to make Claude Code useful at the level of your expertise.

What you’ll learn

Most AI advice says “let AI do the work for you.” But we'll focus on doing the right work, with your judgment, so your expertise compounds.

  • Learn to spot the work where your expertise is the difference, not just what is easy to automate.

  • Avoid getting faster at tasks that never mattered in the first place.

  • Leave with your Leverage Map: the short list of work where you are the edge.

  • Learn what context, parameters, and standards your agent needs to work at your level.

  • Encode your expertise into the system instead of bypassing it.

  • Build your first Judgment System: an agent that holds your standards.

  • Stay in the loop on purpose so your judgment sharpens instead of atrophies.

  • Learn how to review, question, and redirect AI output instead of accepting it passively.

  • Build workflows that keep you intellectually active, not dependent.

  • Turn each workflow pass into reusable knowledge, standards, and examples.

  • Capture your way of thinking so it becomes easier for your agent to use next time.

  • Leave with an asset that gets more valuable every week you keep building.

Learn directly from Vicky

Vicky Zhao

Vicky Zhao

Helping non-technical leaders build framework-powered AI operating system

Previously Worked At/Taught At
Google
L.E.K. Consulting
Oliver Wyman
Morgan Stanley
University of Alberta
See all products from Vicky

Who this course is for

  • AI-forward leaders. You use LLMs, but don't know where to start with agentic workflows.

  • Domain experts and deep thinkers. Your judgment is still stuck in your head. So the AI output you get is currently not good enough to use.

  • The operator, manager, or consultant. You manage too much busywork. You want workflows that free up time for strategic work only you can do.

What's included

Vicky Zhao

Live sessions

Learn directly from Vicky Zhao in a real-time, interactive format.

Async video lessons and setup

Use setup walkthroughs, concept lessons, and docs to get up to speed on your own schedule.

The Agent Stack system

Set up CLAUDE.md, memory files, project structure, and skills so Claude Code works with your expertise.

Frameworks, templates, and prompts

Distill your experience into reusable assets that help your agent produce higher-quality work.

Starter skill library

Install and adapt pre-built skills so you can start from a working base instead of a blank page.

Peer cohort and community

Build alongside other professionals, learn from their examples, and pressure-test your workflows together.

Lifetime session recordings

Rewatch every session whenever you need it. Recordings are uploaded within 24 hours.

Maven Guarantee

Your purchase is backed by the Maven Guarantee.

Course syllabus

9 live sessions • 47 lessons

Week 1

Jul 20—Jul 26

    Agent Fundamentals

    9 items

    Claude Code Fundamentals

    12 items

    What to build?

    5 items

    Jul

    14

    Session #1

    Tue 7/144:00 PM—5:30 PM (UTC)

    Jul

    16

    Session #2

    Thu 7/1610:00 PM—11:30 PM (UTC)

    Jul

    20

    Coworking Session

    Mon 7/2012:00 AM—1:00 AM (UTC)

Week 2

Jul 27—Aug 2

    Creating Agentic Skills

    5 items

    Loops & Routines

    6 items

    Jul

    21

    Session #3

    Tue 7/214:00 PM—5:30 PM (UTC)

    Jul

    23

    Session #4

    Thu 7/2310:00 PM—11:30 PM (UTC)

    Jul

    27

    Coworking Session

    Mon 7/2712:00 AM—1:00 AM (UTC)

Schedule

Live sessions

4 hrs / week

    • Tue, Jul 14

      4:00 PM—5:30 PM (UTC)

    • Thu, Jul 16

      10:00 PM—11:30 PM (UTC)

    • Mon, Jul 20

      12:00 AM—1:00 AM (UTC)

Projects

1 hr / week

Async content

1-3 hrs / week

Testimonials

  • I went in with high expectations for CCW, and it honestly exceeded them. Before this, Claude Code felt like a black box; now it feels like a real thought partner. The turning point was Vicky’s framework for exploring problems laterally and going deeper than the initial problem statement. The mix of pre-recorded sessions, daily nudges, live discussions, and community made the experience genuinely joyous. I now use Claude Code to build PowerPoint slides and visuals that reflect my taste and standards from 15 years in finance and tech. More importantly, it helps me go deeper on insights I normally would not have had time to think through. Go in with a beginner’s mindset. Don’t just learn CC mechanics. Treat it as upgrading your brain’s software. Big shift: I think in outcomes now, not inputs.

    Testimonial author image

    Deepak Talwar

    Director, Corporate Technology
  • CCW helped me move from disorganized notes to a working personal knowledge system. Before, I kept circling the same files; now I can sort, structure, and use what I know. I’d recommend it to anyone with too many ideas and not enough momentum.

    Testimonial author image

    Feenix

    Project Manager
  • Vicky demystifies AI. Information about AI piles up and overwhelms me. Where do you start? Who do you follow? Who do you learn from? True to her philosophy, Vicky distills it to the essentials. Before this cohort, I used AI like Google search. Learning the Claude Code fundamentals changed everything. I was able to start a foundation that helped put everything else into place. I was unable to attend all the sessions. Still, my knowledge exploded. While the cohort uses Claude, the core knowledge and skills transfer to another AI tool. I haven’t finished the course yet, but I’m learning so much. If you use AI daily but don’t truly understand it, take this course. Good stuff!

    Testimonial author image

    Daniel Ng

    Director, NDTV Technical Operations
  • I had never worked with Claude Code before this course and wasn’t sure where to start because there was so much information online. I had taken several of Vicky’s courses before, so I knew this would be value for money. The short videos and notes were a helpful primer and clarified the concepts before live class, where I could try Claude Code on real work. I had listed several use cases before enrolling and was able to build some of them during the course. More importantly, this course gave me a clear path forward to keep trying, learning, and improving. Now I can confidently talk about using Claude Code to solve business problems at work. If you don’t know where to start, or you’re tinkering but struggling to create quality output, this course will give you a real head start.

    Testimonial author image

    Suraj

    Senior Manager
  • A week ago, I wouldn’t touch the Claude Code button. Today I’m code-curious: eager to try, fail, and learn. What changed wasn’t my technical skill. It was finally getting my own expertise out of my head. In change management, facilitation, leadership development, and coaching, our magic touch is often felt but not explicit. Vicky’s framework helped me articulate what I actually do, see, and notice. Then Claude Code gave that expertise a place to live and compound through skills, md files, and routines. AI now frees me to do what I love most: be fully present with clients, while Claude Code helps with notes, organizing, and synthesis. You don’t need to be technical. You need to be curious and willing to let your expertise become something playful, meaningful, and yours.

    Testimonial author image

    Yvonne Yam

    Change Facilitator and Executive Coach
  • Great Teacher, great group of people in the cohort, and great introduction to Claude Code. Opened a great range of possibilities for me.

    Testimonial author image

    Bernie

    Self Employed

3 weeks. 3 real workflows.

1/ Workflow 1: Your first useful workflow.

Stop floating in possibilities. Build a mental map. Choose one real workflow and make Claude Code useful on it.

2/ Workflow 2: Extract your judgment.

Choose a judgment-heavy workflow. Get your expertise, standards, taste, and decision criteria out of your head and into files and skills Claude Code can use.

3/ Workflow 3: Make it compound.

Choose a repeatable workflow. Turn it into a routine, verification loop, app, tool, or roadmap you can keep building on.

Student comments from Zoom chat

See what people are saying

Frequently asked questions

Maven for Teams

Reimbursement

Get your company to pay

Everything L&D needs: email template, receipts, and certificate of completion.

Get reimbursed

Private cohort

Run a cohort for your org

A dedicated cohort with a custom schedule and curriculum, tailored to your team.

Book a private cohort