AI MARKETING LAB CLAUDE CODE

Agent Teams

A hands-on manual for marketing students and practitioners. Spawn three AI teammates (Strategist, Copywriter, Researcher) to develop a complete product launch campaign in parallel. Learn to brief teams, manage coordination, and critique AI-generated work.

Product
IE-Nergy
Duration
2–4 hours
Audience
Marketing, marketing students
Deliverable
Campaign brief

What you'll learn

Learn to structure marketing briefs for parallel AI work. Discover how to define specialist roles (Strategist, Copywriter, Researcher) that produce sharp, specific output instead of generic work. Watch three Claude instances coordinate on the same project—sharing findings, debating ideas, and building on each other's work. Develop judgment for evaluating AI-generated marketing with a marketer's eye.

A
Introduction

Orchestrate Teams of Claude Code Sessions

Coordinate multiple Claude Code instances working together as a team, with shared tasks, inter-agent messaging, and centralized management.

What are Agent Teams?

Agent Teams let you coordinate multiple Claude Code instances working together. One session acts as the team lead, coordinating work, assigning tasks, and synthesizing results. Teammates work independently, each in its own context window, and communicate directly with each other.

Unlike subagents, which run within a single session and can only report back to the main agent, you can also interact with individual teammates directly without going through the lead.

⚠️

Experimental feature — enable before use

Agent teams are experimental and disabled by default. Enable them by adding CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1 to your settings.json or environment. Agent teams have known limitations around session resumption, task coordination, and shutdown behavior.

Requirements: Claude Code v2.1.32 or later. Check your version with claude --version

When to use Agent Teams

Agent teams are most effective for tasks where parallel exploration adds real value. The strongest use cases are:

1
Research and review
Multiple teammates investigate different aspects simultaneously, then share and challenge each other's findings
2
New modules or features
Teammates each own a separate piece without stepping on each other
3
Debugging with competing hypotheses
Teammates test different theories in parallel and converge on the answer faster
4
Cross-layer coordination
Changes spanning frontend, backend, and tests, each owned by a different teammate
📢

For marketing: parallel campaign development

Teammates can investigate different competitors simultaneously, develop positioning in parallel with content creation, or explore different audience angles at once. The strategist's findings directly inform the copywriter's tone. The researcher's gaps shape the strategist's positioning. This feedback loop is where Agent Teams excel.

Agent Teams vs Subagents

Both let you parallelize work, but they operate differently. Choose based on whether your workers need to communicate:

Subagents vs Agent Teams comparison diagram

Comparison table

Aspect Subagents Agent Teams
Context Own window; results return to caller Own window; fully independent
Communication Report back to main agent only Message each other directly
Coordination Main agent manages all work Shared task list with self-coordination
Best for Focused tasks (result matters) Complex work (discussion matters)
Token cost Lower: results summarized back Higher: each teammate is separate instance

Use subagents when you need quick, focused workers that report back. Use Agent Teams when teammates need to share findings, challenge each other, and coordinate on their own.

Why Agent Teams for this exercise

Your campaign strategist, copywriter, and researcher need to talk to each other. The strategist's positioning informs the copywriter's tone. The researcher's competitive findings shape the strategist's recommendations. The copywriter asks clarifying questions about positioning. This back-and-forth collaboration is exactly what Agent Teams enable.

If you only needed three independent first-draft taglines with no interdependence, subagents would be cheaper and faster. But this exercise requires real coordination.

What you'll learn in this lab

By the end, you'll understand:

How to structure a marketing brief
Specificity in audience, constraints, and deliverables directly improves output quality
How to define specialist roles
Vivid role definitions (not generic tasks) produce sharper, less generic work
How to watch AI specialists coordinate
See positioning inform copywriting, research findings shape strategy, and teammates challenge each other's ideas
How to critique AI-generated marketing
Develop judgment to spot generic phrases, missed audience nuances, and unsupported claims
01
Getting started

Prerequisites

~5 min · verification only

Before you start, verify you have these tools installed and accessible on your system:

01
Claude Code 2.1.32+

Check with: claude --version

If you don't have it, install from code.claude.com

02
Node.js 18+

Check with: node --version

If needed, install from nodejs.org

03
tmux (any recent version)

Check with: tmux -V

macOS: brew install tmux · Linux: sudo apt install tmux

04
A system terminal

Terminal.app (macOS), iTerm2, or Linux terminal — NOT VS Code's integrated terminal

✓ Checkpoint
  • All tools installed? Run the version checks above. You should see version numbers for all three.
  • Ready to proceed? Move to Part 02 for installation (if needed) or Part 04 to start the exercise.
04
The exercise

Build a Campaign

~45 min · team work + review

You'll spawn three AI teammates to develop a launch campaign for IE-Nergy, a new energy drink positioned for students and young professionals who need calm, sustained focus—not aggressive hype.

The brief

Key facts about the product:

🥤

IE-Nergy: A New Energy Drink

  • Formulation: 150mg natural caffeine (green tea + guarana), L-theanine, B-vitamins, zero sugar, 15 calories
  • Flavors: Yuzu Citrus, Wild Berry, Matcha Mint
  • Price: €2.50 per can
  • Launch: Madrid & Barcelona first, expanding across Europe
  • Differentiator: Calm focus, not aggressive stimulation. A counterpoint to Red Bull/Monster/Bang
  • Timeline: Six weeks until launch

Your team

A
The Strategist

A senior brand strategist with 10+ years launching challenger brands. Thinks in terms of positioning, jobs-to-be-done, defensible differentiation. Skeptical of generic claims. Has a strong point of view on what NOT to say. Grounds every recommendation in audience insight.

B
The Copywriter

A direct-response copywriter who's written campaigns for Gen Z brands. Writes conversational, specific voice—never corporate. Writes differently for each channel (Instagram ≠ TikTok ≠ email ≠ search). Hates buzzwords like "elevate," "fuel," "unlock."

C
The Researcher

A competitive intelligence analyst. Investigates not just what competitors say, but what they DON'T say—looking for whitespace and positioning gaps. Reports findings as actionable insight, not feature tables.

How to run the exercise

01
Create workspace
mkdir ~/Documents/ienergy-launch cd ~/Documents/ienergy-launch
02
Start tmux session
tmux new -s marketing
You'll see a colored status bar at the bottom.
03
Launch Claude Code with Agent Teams enabled
CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1 claude
This activates the experimental feature just for this session.
04
Spawn your three teammates
Read brief.md, then create an agent team with 3 teammates in tmux split-pane mode. Spawn them with these roles: TEAMMATE 1 — STRATEGIST You are a senior brand strategist with 10+ years launching consumer challenger brands. Think in terms of category positioning, jobs-to-be-done, defensible differentiation. Skeptical of generic positioning. Strong point of view on what NOT to say. Ground every recommendation in audience insight. Deliverable: strategy.md with (1) one-sentence positioning, (2) top 3 messages, (3) recommended channel mix, (4) what we will NOT say and why. TEAMMATE 2 — COPYWRITER You are a direct-response copywriter who's written for Gen Z/Millennial brands. Write conversational, specific voice—never corporate. Write differently for Instagram vs TikTok vs email vs search. Hate buzzwords like "elevate," "unlock," "fuel." Deliverable: copy.md with (1) Instagram caption (125 words), (2) TikTok script (30-second video), (3) email (subject + opening), (4) Google Search ad. Wait for Strategist's positioning before drafting. TEAMMATE 3 — RESEARCHER You are a competitive intelligence analyst. Investigate not just what competitors say, but what they DON'T say. Look for whitespace and positioning gaps. Report as actionable insight, not feature tables. Deliverable: research.md with (1) profile of 3 competitors (Red Bull, Monster, one challenger like Celsius/Tenzing), (2) each competitor's core positioning in one sentence, (3) 2-3 positioning gaps IE-Nergy could own. WORKFLOW: 1. Researcher works first, shares findings 2. Strategist uses research to develop positioning 3. Copywriter waits for positioning, then drafts creative 4. Lead synthesizes all three into campaign-summary.md Have teammates message each other directly when they need input.
As Claude spawns each teammate, tmux automatically splits into multiple panes.

While your team works

Use these shortcuts to navigate and steer:

Command What it does
Ctrl+B → arrow keys Move between panes
Ctrl+Bz Zoom in/out of one pane
Click into a pane → type Send feedback to that teammate
💡

Steering tips

To the Copywriter: "The TikTok script doesn't sound like a TikTok—it reads like a TV spot with shorter sentences. Rewrite it as if explaining to a friend."

To the Strategist: "Focus on what makes IE-Nergy defensibly different from Red Bull. What positioning would they NEVER claim?"

To the Researcher: "Also analyze Celsius and Tenzing—they're the nearest competitors in the functional energy space."

Learning Path

1. Patterns → 2. Concepts → 3. Subagents → 4. Agent Teams (you are here)

Step 1 — Patterns
Blocks, Workflows & Agents
Composable patterns for building agentic systems
Step 2 — Concepts
Intro to Subagents & Teams
When to use subagents vs agent teams
Step 3 — Subagents
Subagents Marketing Lab
Quick guide to subagents in VS Code