Internal Training

Claude Code Training

Marketing Team session. A mental model for working with AI tools, plus the practical setup to start using one today.

Riccardo Vandra Founder, Uplifted Consulting

Why this session

My goal today.

By the end of the session, three things should be true for you.

01

A mental model for how to think about AI tools like Claude Code, so you stop guessing what works.

02

A repeatable way to spot your own use cases so you are not waiting for someone to hand them to you.

03

Onboarded onto the tool yourself ready to use it to speed up the slow work, remove the boring work, and free up time for the work you actually want to do.

The setup

Think back to your first week at any job.

You could not really ship anything until three things were in place.

Show the ropes

Someone telling you what to do and how.

Context

About the company, the team, the audience.

Access

Logins to the systems where the work happens.

AI is the same. Most people skip these three, and that is why they can not get it to actually use their systems. Treat AI like your own intern. Or the version of you on your first week.

The mental model

Every teammate, human or AI, needs three things.

We call them ACT. The whole rest of this session lives inside these three letters.

A

Abilities

The things they can do. Pull a report. Draft an email. Run a segment analysis. The skills they bring to the role.

C

Context

What they know about us. ICP, brand voice, segments, what good looks like.

T

Tools

The systems they can touch. Iterable, Tableau, Figma, Sheets, inbox. The platforms where the work actually happens.

The lesson

Anything you would onboard a person to do, you can onboard Claude to do.

The procedure they would learn becomes a skill. The docs they would reference become context. The systems they would use become tools.

Stack a few of those in one folder and you have a small team running inside it.

Building block 01

Abilities = Skills.

In Claude Code, abilities are called Skills. A skill is one task you would normally teach a teammate.

Pull weekly report

Export from the dashboard, summarise the deltas, drop it into a deck.

Draft a brand-aligned email

Take a brief, write a draft that sounds like us, generate subject line variants.

Analyse a segment

Pull the segment's behaviour over a window, surface the patterns worth acting on.

Skills can call other skills. A bigger skill is just a checklist of smaller ones.

Building block 02

Context.

What the teammate needs to know to do the job well. Three flavours.

Foundational

Who you are. Brand voice, ICP, segments, positioning. Broad background that shapes every decision.

Work-specific

The info one skill needs to run. For "draft an email," that is the campaign brief and the audience segment.

References

Frameworks, templates, past examples. What good looks like.

Building block 03

Tools.

How Claude reaches your systems. Three kinds of interface.

API

Every action a platform exposes. Direct, no UI. Best when the platform has full docs.

CLI

A packaged subset of an API. Easier for Claude to discover and use. Faster to set up.

MCP

Best for tools that need third-party login (Google, etc.), or when you want a record of what Claude did.

Examples you would plug in: Iterable, Tableau, Figma, Sheets, Gmail.

Building block 04

Skills, context, and tools all live in one folder.

That folder is the workspace. Open it in VS Code and Claude operates inside it.

marketing-workspace/
│
├── CLAUDE.md                          // main instructions
│
├── .claude/skills/                    // abilities
│   ├── pull-weekly-report/
│   ├── draft-brand-email/
│   ├── analyse-segment/
│   ├── iterable/                     // tool integration
│   └── tableau/
│
├── context/                           // brand, segments, tone, ICP
│
└── workspace/                         // where the actual work happens
    └── campaigns/q2-reengagement/
        ├── brief.md
        ├── draft.md
        └── report.md

In motion

What happens when you call a skill.

You say: "Draft a re-engagement email for lapsed users." Here is what the workspace runs through.

Skill

Reads draft-brand-email/SKILL.md for the procedure.

Context

Reads context/brand/tone.md and context/segments/lapsed-users.md.

Judgment

If the brief is vague ("users we lost"), pushes back for a clearer one.

Tool

Opens Iterable to pull recent similar campaigns. Drafts in the workspace.

Output

Writes draft email and subject line variants into the campaign folder.

Skill, context, and tools, working as one inside the workspace.

Get set up

Three ways to install Claude Code.

Pick the one that fits how you already work. You can switch later.

01

Claude Desktop App

Easiest path. No terminal. Best for trying it out without setup overhead.

02

VS Code + Extension

Recommended for daily work. Claude lives alongside your files and notes.

03

Claude Code CLI

For terminal users who want full control. Same engine, no IDE.

Today we will demo on VS Code + Extension.

Find your own

Identify your own use cases.

You do not have to wait for someone to hand you a list. Two foundational skills exist so the workspace can help you find and build your own.

Explore

Walks you through your week. Asks what drags, what repeats, what you wish was already done. Surfaces a shortlist of tasks worth turning into skills.

Skill builder

Takes one use case you picked and scaffolds it into a working skill. Procedure, context it needs, tools it touches. Ready to run.

We will run Explore live on one person in the room.

In production

Two skills we built for Bark.

Both live in the same workspace. Both follow ACT. One does the analysis, one builds the foundation other skills run on top of.

Google Ads Review

Pulls weekly account performance and writes a review plus a recommended-actions doc.

Iterable Inventory

Snapshots every campaign and template in Iterable. Foundation for downstream skills.

Example 01

Google Ads Review.

Run once a week. Output lands in the workspace as a review doc plus a separate actions doc, in both markdown and HTML.

workspace/google-ads-reviews/
│
├── 2026-05-14.md                  // the review
├── 2026-05-14.html                // shareable version
│
├── 2026-05-14-actions.md          // recommended actions
└── 2026-05-14-actions.html

ACT in one skill: ability (review the account), context (KPI definitions, account structure), tool (Google Ads).

Example 02

Iterable Inventory.

On its own, an inventory is not high-leverage. Its value is as a foundation. Once Claude has a full map of campaigns and templates, other skills become possible on top of it.

Template audit

Spot duplicates, unused blocks, dead campaigns. Cleanup work that was never worth doing by hand.

Variant generation

Take a winning template, generate brand-aligned variants for new segments or tests.

Sheets export

Push the inventory into a sheet so the team can sort, filter, and decide what to retire.

One skill builds the map. The other skills are the journey.

Open floor

Questions.

Ask anything about ACT, the workspace, the examples, the install, or your own setup.

Email riccardo@upliftedconsulting.com

LinkedIn riccardovandra

YouTube @riccardovandra