A Marketer's Guide to Setting Up Claude


Hello Reader

I run a lot of AI workshops with marketing teams.

When I ask who’s using AI tools, every hand goes up. When I ask who has a coordinated team effort behind it, most of the hands come down.

A lot of teams are moving to Claude right now, or switching over to take advantage of its model capabilities and connected tools.

Buying the license is the easy part.

What most teams skip is the setup and coordination that makes the license worth paying for.

Without a shared foundation, you get the same pattern I see everywhere: credit costs ballooning, super users pulling ahead while the rest of the team gets inconsistent results and nobody able to say what the AI investment is producing.

This week I want to walk you through the setup step that gives your whole team a shared starting point before anyone opens a new chat.

How to Set up Claude for Your Team

When your team skips setup, every person is running a different version of Claude. People are cutting and pasting the same prompts at the start of every session, multiple people are solving the same problem and no one’s really sure what all the AI agents are doing.

This means wildly variable output quality, credits spent on unnecessary repetition, and a growing gap between the two or three people with technical skills and everyone else.

According to the Averi AI 2026 benchmarks report, there are four maturity levels that determine content performance (although the same framework applies to all marketing functions):

Level 1

Ad Hoc AI Usage. Using AI tools for one-off tasks without persistent brand context or strategy architecture.

Level 2

Integrated AI Tools. Multiple AI tools connected through manual workflows. SEO platform + AI writing tool + CMS + analytics, operated by a human who serves as the integration layer.

Level 3

AI Engine. Purpose-built platform with persistent brand context, strategic architecture, AI drafting with multi-dimensional scoring, native CMS publishing, and analytics feedback loops.

Level 4

Autonomous Operations. AI agents that proactively create, optimize, publish, and iterate with minimal human oversight.

Averi estimates that 50% of teams are at Level 1, and only 5% have reached Level 4.

In my experience working with marketing teams at L'Oreal, HP and Autodesk, Foundation is the layer that gets skipped most often and costs the most in wasted output.

And you can’t reach Level 4 without laying the groundwork.

Your Five Minute Setup Guide

Most people open Claude and start typing.

The output is generic because Claude has no context about who you are, what your brand sounds like or what you need.

Step 1: Fill in your profile

Go to Settings > Profile. This is where you tell Claude who you are and how you want it to work with you. Write it the way you would brief a senior contractor on their first day.

Here is part of my own profile: "Use rules from Strunk and White's style guide. Do not use sentence fragments for emphasis. Do not narrate your thinking, plan or process..."

Claude applies this to every response and session, without you repeating it. The more specific you are, the less time you spend correcting output.

Step 2: Set up your first Project

Projects are persistent workspaces and one of the best ways to coordinate your team. You can build as many as you need: one per project, campaign or performance reporting.

Start with a Project your entire team can use: a brand guide. Load it with your voice guidelines, approved messaging hierarchy and tone of voice examples, then add a system prompt that tells Claude its job: "You are a brand reviewer for [Company]. When I give you copy, check it against the guidelines in this Project and flag anything that does not match our voice, tone or approved claims."

When anyone on the team opens this Project and pastes in a draft, they get a consistent, brand-informed review. No more emailing the brand team for a quick check.

Step 3: Upload reference materials selectively

Every document you load into Claude is read in full each time it responds, which means uploading everything you've ever written costs you credits on context Claude doesn’t need.

A focused 2-page voice guide produces better output than a 200-page brand bible and costs fewer credits per session to process.

Step 4: Import memory from another tool

If your team has been using ChatGPT (or another LLM), you can import memory via Settings > Capabilities > Import memory from other AI providers. Worth doing if there is useful context built up there.

Build the System

Once you’ve completed your set up you can start using Claude as a system. This means setting up your Skills and Connectors.

Skills

Skills are saved instructions that tell Claude how to behave for a specific task.

A Skill might contain your tone of voice guidelines, a brief format your team uses every week or the research steps for a specific campaign type. Once a Skill is saved, Claude applies it automatically when the task matches.

Skills most useful for marketing teams:

Writing Voice — Upload your brand guidelines or a set of writing rules and Claude will apply them to every piece of copy it produces.

Audience Profile — Describe your buyer, their pain points and the language they respond to, so Claude writes for that person every time.

Campaign Brief Template — Save your standard brief structure and Claude will populate it from a conversation, a call transcript or a set of notes.

Research Workflow — Document the sources, formats and analysis steps you use for competitive or market research and Claude will follow them consistently.

To use Skills: go to the Customize section, select Skills and click the + icon > Browse Skills or Create Skill.

Connectors

Connectors link Claude directly to the platforms your team already uses. Once connected, Claude can read from and act inside those tools without copying and pasting between tabs.

The connectors most useful for marketing teams:

Google Drive — Pull briefs, search documents and reference campaign files without leaving the conversation.

Notion — Read, create and update pages using plain language.

HubSpot — Pull contact records, review campaign performance and draft follow-up sequences without leaving the conversation.

Canva — Give Claude approved copy and tell it to populate a social template directly.

To use Connectors: go to the Customize section, select Connectors and click the + icon > Browse Connectors.

Helpful Hacks

A few things worth knowing before your team starts working inside Claude.

Type /compress in any chat.

This condenses the full conversation history into a compact summary. It keeps your context, reduces the processing load and keeps credit consumption in check on longer sessions.

Understand the three memory levels.

Within a conversation, Claude holds everything discussed and uses it to inform every response. The longer and richer the session, the better the output, which is why continuing an existing conversation produces better results than opening a new one for the same project.

Across conversations, everything loaded into a Project persists in every session your team opens inside it. Claude also generates its own memories from your conversations, which you can view, edit and delete at any time via Settings > Capabilities.

Human-in-the-Loop

Claude generates content based on what you give it. That means the quality of the output depends entirely on the quality of the input and the critical eye you apply before anything leaves your team.

A Marketing Director's Guide to Claude includes all your set up instructions plus six ready-to-run workflows

  • Campaign Brief Analyst
  • Brand Voice Engine
  • Performance Reporter
  • Ad Copy Generator
  • Competitive Research Analyst
  • Audience Builder

Each with the prompts to run them, the files to upload first and the follow-up prompts that get you to a usable output.

Download it free at https://ai-at-work.kit.com/claude.

Want to Level Up Your AI Game?

If your team is ready for a hands-on AI strategy session, my custom-designed workshops are built to uncover the workflows that can save you hours every week.

Prefer to start small? My YouTube channel is packed with quick, practical “how-to” videos that show you exactly how I use AI tools for marketing, content, and automation.

Planning an event or conference? I deliver high-energy AI sessions that engage audiences and leave them with actionable strategies they’ll talk about long after the event. Book me for your event here.

Did some one forward you this email? You can subscribe here.

2120 Contra Costa Blvd #1059 , Pleasant Hill, CA 94523
Unsubscribe · Preferences

AI at Work

AI at Work is a weekly newsletter on how marketing teams redesign workflows, roles, and systems with AI. Real examples, practical frameworks, and repeatable processes operators can use immediately. Join thousands of successful marketing leaders by subscribing below!

Read more from AI at Work
AI Content Creation

Hello Reader Years ago, when I was VP of Marketing at a SaaS platform, I had a content problem I couldn't solve with budget. We were too small to justify a full-time writer and I was already doing the work of three people, so doing it myself wasn't realistic. The answer was freelancers. I'd find a writer, brief them on our product and audience (corporate travel managers, a niche with its own language and priorities) and wait two weeks for something I could publish. What came back was almost...

Hello Reader Your team just spent two weeks building an outbound sequence that generated four replies, two of which were out-of-office messages. If you’re experiencing this scenario then it will come as no surprise that reply rates on B2B email have dropped from 7% to 3.43%, according to Instantly’s 2026 Benchmark Report. At the same time, an estimated 376 billion emails are sent every day, a number expected to climb past 408 billion by 2027. With the math working against you, what are you...

Hello Reader A few weeks ago I was building a Claude workflow to update a product page and I went looking for real opinions on Reddit. I searched three subreddits, found 14 threads and read every reply. People were naming competitors, describing specific frustrations, features they loved and using words and phrases I could use in my copy. It was the most useful 45 minutes of research I’d done in months. Then I asked ChatGPT to recommend a vendor in that category. Two of the threads I’d just...