Leveraging AI

AI is not a standalone topic in this course. It is the way you will apply nearly every part of The Consultant’s OS more effectively.

Starting in Week 2, you’ll learn how to use AI tools to accelerate your research, sharpen your analysis, and improve how you are able to communicate. The OS remains the foundation; AI simply increases your leverage.

Getting Leverage with AI

Think of AI as an amplifier for The Consultant’s OS:

OS Capability	Without AI	With AI
Research	Hours of manual searching	Rapid synthesis across sources
Analysis	Slow iteration on models	Fast hypothesis testing
Communication	Multiple draft cycles	Accelerated refinement
Learning	Delayed feedback	Instant practice partner

AI doesn’t replace judgment, structure, or rigor. It makes each of them faster, cheaper, and more scalable.

You’ll use two broad categories of AI tools in this course. Both rely on the same underlying language models. In principle, either could answer the same question.

The difference isn’t the underlying model but the tools wrapped around it. Web chatbots operate through a browser interface with limited, manual file uploads and limited agentic behavior. AI agents operate through you a code editor and terminal (CLI), with direct access to your file system, the ability to execute code, and the capacity to chain actions together. Same model but different capabilities.

This leads to an emerging discipline you’ll practice throughout the course: context engineering, which is deciding what information an AI system has access to when it responds.

Prompt engineering focuses on how you ask: wording, structure, and tactics for framing a request.
Context engineering focuses on what the AI knows: your files, your data, your project, and your constraints.

You can write a great prompt, but if the AI doesn’t have access to your data, your documents, or know about your project, it can only be so helpful. Context engineering is about optimizing what information and tools the AI model has access to in order to push it’s abilities far beyond what’s possible from a prompt in a web based chatbot.

Web Chatbots

Tools: Claude, ChatGPT, Gemini

Web-based chatbots are amazing, almost magical conversational interfaces. You type; the AI responds; you refine. They live in your browser and require no setup.

You can upload files to chatbots, but that context is manual and fragile. You decide what to upload, conversation by conversation. The AI sees only what you explicitly provide, cannot infer relationships across files, and forgets everything once the conversation ends.

Chatbots are ideal when the AI doesn’t need deep knowledge of your specific project. They shine in situations like:

Research and synthesis — Understanding industries, companies, or unfamiliar concepts
Structuring thinking — Building MECE frameworks, testing hypotheses, developing storylines
Writing and editing — Drafting emails, refining prose, sharpening arguments
Practice and preparation — Case interview drills, networking conversation prep

When you just need to think, explore, or communicate, chatbots are often the fastest option.

AI Agents:

Tools: Claude Code, Gemini CLI, GitHub Copilot, OpenAI Codex, Cursor

AI agents use the same underlying models as web chatbots but they are accessed and operate primarily through a locally installed terminal or code editor such as VS Code, Antigravity, or Cursor.

That is, AI agents are accessed via tools that historically have been the domain of software developers and data scientists, in places where few consultants dared enter.

For most of computing history, these tools required specialized knowledge that tools years to develop. You had to understand syntax, commands, file systems, and obscure error messages. Those knowledge barriers kept them largely confined to developers.

That barrier is now gone.

Large language models are fundamentally text-in, text-out systems. They excel at reading text, generating text, and reasoning over text.

VS Code and the command line (terminal) are also fundamentally text-based:

Source files are text
Data files are often text
Logs and errors are text
Commands are text

This makes them the native habitat of LLMs.

Graphical user interfaces (GUIs) were designed for humans. Because we are vision-dominant, windows, icons, and buttons made computers easier to use. Even the name Windows is a self-reference to this design choice. But what is intuitive for humans is often unnecessary for AI agents, which operate natively in text.

This has led to a “CLI renissance” where the command line is very much back in style and more powerful than ever before.

With AI agents assisting you, tools that once required years of training now allow you to describe what you want in plain English while the AI translates that intent into concrete actions.

This is the quiet revolution that so far few people have put words to or are fully appreciating: developer tools have become general-purpose thinking and execution tools for all knowledge workers.

Because agents operate inside your editor and terminal, they gain something chatbots cannot: full context.

They can:

Read your project files
Understand how documents and data relate
Write new files and modify existing ones
Run code, see errors, and fix them
Chain multi-step workflows end to end

A useful mental model:

A web chatbot knows everything about the world but nothing about your project.

An AI agent knows everything about the world and can see exactly what you’re working on.

That difference matters enormously when you want to get the most out of AI models.

Chatbots vs AI Agents

Criteria	Web Chatbot	AI Agent
Response speed	Faster	Slightly slower (reads files first)
Setup required	None (just a browser)	Installation required
File context	Manual upload, limited	Automatic, full project access
Code execution	Limited (sandboxed Python)	Full (any language, your machine)
Output persists	No (copy-paste)	Yes (writes to files)
Best for	Thinking and drafting	Building and executing

When to Use Each

If you need to…	Web Chatbot	AI Agent
Brainstorm ideas	✓	✓
Research a company	✓	✓
Structure an argument	✓	✓
Practice a case interview	✓
Analyze your dataset		✓
Build a model from your data		✓
Create files or deliverables		✓
Automate a repetitive task		✓

Rule of thumb: if the AI needs to take action, use an agent. Otherwise, use whatever is fastest.

What counts as “action”? Anything beyond generating text:

Reading your files — analyzing datasets, reviewing documents, understanding your project
Writing files — creating deliverables, saving outputs, building persistent artifacts
Running code — executing scripts, testing solutions, iterating on errors
Searching your system — finding files, grepping through codebases, locating specific content
Executing commands — installing packages, running builds, interacting with APIs
Chaining operations — multi-step workflows where each step depends on the last

If your task involves any of these, reach for an agent. If you just need to think through a problem or get information, a chatbot in your browser is faster and simpler.

Setting Up Your AI Workflow

The Tools

VS Code is the open-source code editor (often called an IDE for Integrated Development Environment) that serves as the foundation for many AI coding tools. Both Cursor and Antigravity are built on top of VS Code, adding AI capabilities while preserving the familiar interface. If you learn one, you can use the others.

Recommended setup for this course:

Antigravity (Google) — Our primary tool due to its generous free tier
Claude Code — Recommended for students who want the most capable agentic coding experience ($20/mo). You can access Claude Code from any of VS Code, Antigravity, or Cursor, all of which have a very simliar look and feel. If you know one, you know them all.

AI Coding Agents Comparison

Company	Product	Student Cost	CLI	VS Code Extension	Dedicated IDE	Web
Anthropic	Claude Code	$20/mo	✓	✓		claude.ai/code
Google	Gemini	Free Tier	✓	✓	Antigravity	aistudio.google.com
Microsoft	GitHub Copilot	Free Tier	✓	✓	VS Code	github.com/copilot
OpenAI	Codex	$20/mo	✓	✓		chatgpt.com/codex
Cursor	Cursor	Free Tier	✓		Cursor	cursor.com/agents

Step 1: Install Software

Download and install the following IDEs on your laptop. Overtime you will tend to use only one but good to have exposure to all of them.

Step 2: Configure Your Environment

Tip

Troubleshooting tip: If at any point you get stuck or something is not working as expected, take a screenshot or copy/paste the command you ran plus error text into ChatGPT, Claude, or Gemini and ask for help troubleshooting.

Mac

Open VS Code and launch a new terminal (Terminal → New Terminal). The terminal should appear as a new window pane within VS Code.

Copy/paste the following into the terminal and hit enter

# Download and install nvm:
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.3/install.sh | bash

# in lieu of restarting the shell
\. "$HOME/.nvm/nvm.sh"

# Download and install Node.js:
nvm install 24

# Verify the Node.js version:
node -v # Should print "v24.13.0".

# Verify npm version:
npm -v # Should print "11.6.2".

Once complete you are ready to install the AI agents by running the following commands in the terminal.

Claude Code:

curl -fsSL https://claude.ai/install.sh | bash

After installation, run the following command in the terminal to activate the agent. (requires a paid Claude account, $20 per month)

claude

Google Gemini CLI:

npm install -g @google/gemini-cli

After installation, run the following command in the terminal to activate the agent.

gemini

GitHub Copilot:

npm install -g @github/copilot
copilot

After installation, run the following command in the terminal to activate the agent.

copilot

OpenAI Codex:

npm install -g @openai/codex
codex

After installation, run the following command in the terminal to activate the agent.

codex

Windows

Open PowerShell as Administrator:

Press Start
Type PowerShell
Right-click Windows PowerShell
Select Run as administrator

Copy/paste the following into PowerShell and hit enter:

# Download and install fnm (Fast Node Manager):
winget install Schniz.fnm

Close and reopen PowerShell (as Administrator), then run:

# Download and install Node.js:
fnm install 24

# Verify the Node.js version:
node -v

# Verify npm version:
npm -v

Once complete you are ready to install the AI agents by running the following commands in the terminal.

Claude Code:

irm https://claude.ai/install.ps1 | iex

After installation, run the following command in the terminal to activate the agent. (requires a paid Claude account, $20 per month)

claude

Google Gemini CLI:

npm install -g @google/gemini-cli

After installation, run the following command in the terminal to activate the agent.

gemini

GitHub Copilot:

npm install -g @github/copilot

After installation, run the following command in the terminal to activate the agent.

copilot

OpenAI Codex:

npm install -g @openai/codex

After installation, run the following command in the terminal to activate the agent.

codex

Step 3: Practice the Basics

Try these starter exercises:

Research synthesis: “What are the key trends affecting [your target company’s] industry?”
Structure building: “Help me create a MECE issue tree for [problem]”
Case practice: “Give me a market sizing question and walk through it with me”

Session 2 Activity: AI as Your Operating Amplifier

Recruiting: AI-powered networking prep

Use AI to prepare for networking conversations by researching the person and firm before you reach out.

Help me prepare for a networking conversation with a consultant:

**Person I'm Meeting**
- Name: [NAME]
- Firm: [FIRM NAME]
- Role: [TITLE]
- LinkedIn profile: [URL or key details]

**Prep Questions**
1. What does this person's career path tell me about what they value?
2. What are 3 thoughtful questions I could ask that show I've done my homework?
3. What aspects of my background might resonate with their experience?
4. What's happening at their firm right now that I should know about?

**Conversation Goals**
Help me identify:
- 1-2 specific things I want to learn from this conversation
- How I can be helpful or interesting to them (not just ask for things)
- A natural way to follow up after the conversation

Client Work: Intro to Claude Code: 10-K analysis

Learn to use AI to analyze financial filings and extract key insights from your target company’s 10-K.

Help me analyze the 10-K filing for [COMPANY NAME]:

**Company Overview**
1. What does the company describe as its core business?
2. What are the key segments and what % of revenue does each represent?
3. What does management highlight as key competitive advantages?

**Financial Analysis**
4. What are the 3-year trends in revenue, gross margin, and operating margin?
5. What are the largest expense categories and how are they trending?
6. What does the cash flow statement tell us about the business model?

**Risk Factors**
7. What risks does management highlight as most significant?
8. Are there any risks that seem understated or missing?

**Management Discussion**
9. What is management's narrative about the company's performance?
10. What strategic priorities do they emphasize for the coming year?

Summarize the key takeaways in 3-5 bullet points.

AI-Assisted Consulting Actions

Think Clearly with AI

Core Action	How AI Helps
Diagnose current state	Synthesize news, filings, analyst reports quickly
Define the problem	Pressure-test your problem statement for clarity
Build MECE structures	Generate and critique issue trees

Get to the Right Answer with AI

Core Action	How AI Helps
Build fact base	Research companies, industries, competitors rapidly
Estimate and bound	Sanity-check assumptions, find benchmarks
Synthesize	Draft “so what” statements, identify patterns

Move Work Forward with AI

Core Action	How AI Helps
Workplanning	Draft project plans, identify dependencies
Progress without certainty	Explore multiple scenarios quickly

Create Impact with AI

Core Action	How AI Helps
Craft storylines	Structure Pyramid Principle arguments
Communicate with brevity	Edit for conciseness, strengthen logic
Prepare for stakeholders	Anticipate questions, practice responses

Session 3 Activity: AI Workflows & Team Norms

Recruiting: Build detailed firm research profile

Use AI to build a comprehensive research profile on your target firms, going deeper than surface-level information.

Help me build a detailed research profile for [FIRM NAME]:

**Firm Deep Dive**
1. What is their official value proposition and how do they differentiate from competitors?
2. What are their key practice areas and industry focuses?
3. What notable projects or client work have they publicized recently?
4. Who are their key leaders and what backgrounds do they have?

**Culture and People**
5. What do Glassdoor/Fishbowl reviews say about the culture (patterns, not outliers)?
6. What do LinkedIn profiles of recent hires tell us about who they value?
7. What do current employees post about on LinkedIn?

**Recruiting Intelligence**
8. When do they typically recruit? What's their timeline?
9. What's their interview process? How many rounds?
10. What types of cases do they tend to give?

**My Fit Assessment**
Based on all this research:
- Where am I a strong fit?
- Where do I have gaps to address?
- What should I emphasize in conversations with this firm?

Compile into a 2-page firm profile I can reference.

Client Work: Create structured company analysis template

Use AI to create a reusable template for analyzing companies throughout your project.

Help me create a structured analysis template for [COMPANY NAME] that I can use throughout the semester:

**Company Profile Template**
Create sections for:

1. **Company Overview**
   - Business description
   - Key products/services
   - Geographic footprint
   - Ownership structure

2. **Financial Summary**
   - Revenue and growth trends (5-year)
   - Profitability metrics
   - Key financial ratios
   - Capital structure

3. **Strategic Position**
   - Market position and share
   - Key competitors
   - Competitive advantages
   - Strategic challenges

4. **Industry Context**
   - Market size and growth
   - Key trends
   - Regulatory environment
   - Disruption risks

5. **Management and Governance**
   - Key executives
   - Recent strategic decisions
   - Stated priorities

6. **Investment Thesis / Key Questions**
   - Bull case
   - Bear case
   - Key uncertainties to resolve

Make this a template I can populate and update as I gather more information.

AI Ethics and Professional Standards

What’s Acceptable

Using AI to accelerate research you would do anyway
Getting feedback on your writing and logic
Practicing cases with AI as a partner
Generating first drafts that you then refine

What’s Not Acceptable

Submitting AI-generated work as entirely your own without disclosure
Using AI during actual interviews (unless explicitly permitted)
Relying on AI without developing your own judgment

The 80/20 Rule

AI can get you 80% of the way quickly. The last 20%, including judgment, nuance, and client context, is where you add value. Never outsource your thinking entirely.

Getting Started This Week

By the end of Week 2, you should be able to:

Access Claude Code or Gemini within VS Code, Anitgravity, or Cursor
Use an AI agent to help you conduct research and save the output to a markdown file for future reference.

Come to class ready to demonstrate your setup.