Part 1 — The Hire: Onboarding Your AI VP

Part 1 of a 7-part series on building your AI VP of Marketing with Claude Cowork. Next: Part 2 — The Playbook: Rules and Standards

Why an AI VP of Marketing

You spend your evenings writing Instagram captions, your weekends wrestling with email automation, and your mornings wondering if any of it moves the needle. Marketing keeps falling to you because a real VP costs six figures and your business is not there yet. So the work gets done in the gaps — without strategy, without positioning, without anyone whose job is to think about the big picture.

A chatbot does not fix that. Ask it to "write a social media post about our new product" and you get a post. Fine. But a VP thinks about questions you have not asked yet — tells you to drop Twitter and double down on LinkedIn because that is where your buyers are, notices your email open rates slipping before you check the dashboard, remembers that your last launch flopped because you announced it on a Friday afternoon. This series gives you that VP, built on Claude Cowork, running on your desktop, working from a brief you write together.

Claude Cowork is a desktop application from Anthropic — not a chatbot, not a browser tab you close and forget. Unlike ChatGPT or Claude.ai, where conversations fade from memory, Cowork lets you create a Project with persistent instructions that carry across days and weeks. Your AI VP does not forget your brand voice between Tuesday and Thursday. Cowork can also work autonomously — hand it a task, step away, come back to finished work. That matters for marketing, where the job is often "produce five variations of this email subject line" or "draft next week's social calendar."

Prerequisites: a Claude Cowork subscription and a business to market. If you can describe what you sell and who buys it, you have everything you need.

Create Your Cowork Project

Open Claude Cowork. From the home screen, click New Project. The Project you create here is where everything in this series will live.

Give your project a name. "Marketing VP" works. So does your company name followed by "Marketing." The name is for you — pick something you will recognize when you have multiple projects open.

You should see an empty project workspace. No files, no instructions, no conversation history. This is day one. Your VP just walked into an empty office with a blank whiteboard and no context about the company.

The Onboarding Interview

Every good hire starts with onboarding. Your VP needs to understand your business before they can market it. Instead of writing a brief from scratch, you are going to let your VP interview you.

The process works like this: you paste a prompt into Cowork, and your VP asks you a series of questions about your business. You answer each one. At the end, the VP compiles your answers into a structured briefing document called CLAUDE.md. This document becomes the foundation for everything that follows.

Copy the following prompt and paste it into your new Cowork project:

You are my new VP of Marketing. Before you do any work, you need to understand my business. Interview me to build your briefing document.

Ask me the following questions ONE AT A TIME. Wait for my answer before moving to the next question. Be conversational -- if my answer is vague, ask a follow-up to get specifics.

Questions to cover:
1. What is the name of your business?
2. What do you sell or offer? (Be specific -- not just "consulting" but what kind and for whom)
3. Who is your ideal customer? (Demographics, role, industry, company size -- whatever applies)
4. How would you describe your brand personality in 3-5 adjectives? (If you are not sure, describe how you want customers to feel when they interact with your brand)
5. What makes you different from competitors? Why do customers choose you over alternatives?
6. What are your top 3 marketing goals right now? (Examples: more leads, higher website traffic, better email engagement, launch a new product)
7. What marketing channels are we currently using, if any? (Social media, email, blog, ads, word of mouth, etc.)
8. What is working? What is not?

After I have answered every question, move through the four phases below in strict order.

PHASE 1 -- ANNOUNCE FILES
Before you draft anything, name the files and stores you plan to touch.
Name the CLAUDE.md path at the Project root, plus any Memory entries you plan to add.
Do not write a single byte until I have seen this list and said to proceed.

PHASE 2 -- PAUSE FOR WRITES
Draft the full CLAUDE.md in the conversation using clear markdown headers for each required section.
Include Business Overview, Target Audience, Brand Voice, Value Proposition, Marketing Goals, and Current Channels.
Start the draft with this heading line: # Marketing VP -- [Company Name].
Follow it with: This is the operating brief for all marketing work in this project.
Show me the complete draft and then stop and wait for my review.

PHASE 3 -- COMMIT ON APPROVAL
Do not save the CLAUDE.md to the Project until I reply with a clear yes.
If I request changes, revise the draft and show it again before saving.
The only trigger for a write is my explicit approval in this chat.

PHASE 4 -- CLOSE WITH LOCATION
Once the save is complete, confirm the write back to me in one short line.
Use this exact phrasing: here is where it lives: ./CLAUDE.md at the Project root.
If the save failed or landed elsewhere, tell me that instead of guessing.

Begin the interview now with question one.

After PHASE 4 confirms the write, your Project folder surfaces exactly one file at the root: ./CLAUDE.md.

The four-phase protocol you just pasted is the shape every prompt in this series takes. PHASE 1 tells you what the VP is about to touch. PHASE 2 shows you the draft. PHASE 3 waits for your yes. PHASE 4 closes with the file path so you can verify the write. The line here is where it lives: ./CLAUDE.md at the Project root is the signature you will see after every approved write in this guide. If it ever goes missing or points somewhere unexpected, stop and investigate.

Answer the interview honestly and specifically. "We sell software" is less useful than "we sell inventory management software for independent bookstores." The more specific you are, the better your VP performs on every task that follows.

A few tips for the questions that stump people: on brand personality, think about a real interaction with a happy customer — warm and casual? Direct and efficient? Playful? Those feelings are your voice. On marketing goals, be honest — "I want more leads" is fine; "I want to go viral" is a wish. On what is working: if you have no idea, say so. Your VP needs to know what to keep, what to fix, and what to stop.

The interview takes five to ten minutes.

Here is what a finished CLAUDE.md might look like for a fictional flat-rate bookkeeping service called Tideway:

# Marketing VP -- Tideway Bookkeeping

This is the operating brief for all marketing work in this project.

## Business Overview
Cloud-based bookkeeping for freelancers and solo consultants. Flat monthly
fee for categorized transactions, monthly reports, and tax-ready files.

## Target Audience
Freelance designers, developers, writers, consultants earning $75K-$250K.
Skilled at their craft, overwhelmed by financial admin. Ages 28-45, remote,
tried DIY books before giving up.

## Brand Voice
Adjectives: friendly, plain-spoken, reassuring, efficient. A smart friend
who happens to be great with numbers. No jargon, no condescension.

## Value Proposition
Flat-rate pricing with no surprises. Competitors charge by the hour or by
transaction volume. Tideway charges one price regardless of volume.

## Marketing Goals
1. Grow email list from 400 to 2,000 subscribers by Q4
2. Publish 2 blog posts per month on "freelance bookkeeping" keywords
3. Launch a referral program generating 10 new clients per month

## Current Channels
- Blog (2 posts/month, best-performing channel)
- Instagram (3x/week, low engagement)
- Email newsletter (monthly, 38% open rate)
- Word of mouth (strongest but unpredictable)

Your version will look different. Structure matters more than length. Once you are satisfied with the briefing document, save it. Your VP is onboarded.

Where Everything Lives

Your VP does not float in the void — the Project is the office, and a few labelled surfaces are where the company handbook, the filing cabinet, and the house rules live.

A Project is the top-level container for everything your VP needs: the instructions you write, the memory Cowork keeps, and the conversation history that accumulates as you work. Every surface you will touch in this series lives inside one Project, and nothing inside a Project leaks to any other Project you create.

• CLAUDE.md — the briefing document you author. It is the Anthropic-standard filename for project-level instructions, and Cowork loads it automatically at the start of every conversation. You write it; you edit it; you own it.
• Memory — the running record of what your VP has learned about your business. Cowork manages this surface for you, summarizing prior conversations and exposing them in the Project sidebar so you do not have to re-explain a decision you already made.
• Rules files — optional, modular instructions you author for specific situations (a voice rule, a publishing checklist, a brand-safety guardrail). You write them; Cowork loads the ones that apply when they apply. Part 2 is where Rules get their own folder.

Project metadata and Memory are Cowork's job. CLAUDE.md and Rules are yours. Treat the Project as your VP's office and treat these files as the documents on the desk.

The First Assignment

Your VP of Marketing has a desk and a name badge. They know your company name, your industry, your brand voice, your target audience, and your top three goals. That is roughly what a new hire knows after reading the offer letter and skimming the company website.

No executive produces useful strategy from that alone. Before you hand over a real assignment, your VP needs the context that separates generic advice from advice that fits your business. Who are you competing against? What marketing have you tried? What is actually broken right now? Along the way, you will meet Cowork's Memory feature, which means the VP retains what it learns and gets sharper with every conversation.

Deepening the Brief

The CLAUDE.md you built above covers the essentials. Strategy requires a different layer — the kind that lives in your head and rarely makes it into a document. Your VP needs to understand five things before it can produce anything useful:

• Competitive environment. Top three to five competitors, what they do well, where they are weak. If you sell accounting software for freelancers, your VP needs to know that Competitor A dominates on pricing while Competitor B owns the "easy to use" message.
• Current channels. What marketing you are actually doing today — social accounts, email lists, paid ads, referral programs — and honestly, which ones are working and which are running on autopilot.
• Pain points. The specific problems keeping you up at night. Not "we need more leads" — but leads from where, for what, and why the current approach is failing.
• Budget and capacity. Both money and time. A strategy that assumes a designer, a copywriter, and ten hours a week is useless if you have four hundred dollars a month and Tuesday afternoons.
• Timeline pressures. A product launch in six weeks, a seasonal sales window, a conference in August — anything that should shape what gets prioritized first.

You could type all of this into CLAUDE.md directly. Faster approach: let the VP interview you. Paste the following into your Cowork project and answer the questions it asks.

I need you to interview me so you can build a complete picture of our marketing situation. Ask me these questions one at a time -- wait for my answer before moving to the next one.

1. Who are our top 3-5 competitors? For each one, what do they do better than us at marketing, and where are they weak?
2. What marketing channels are we currently using? For each one, is it working, underperforming, or something we just set up?
3. What are the 2-3 biggest marketing problems we face right now? Be specific -- not "we need more leads" but what exactly is broken.
4. What is our realistic monthly marketing budget? Include both money and time (hours per week you can personally spend on marketing).
5. Are there any upcoming deadlines, launches, or events that should shape our priorities in the next 90 days?

After I answer all five, summarize what you learned and add it to your knowledge of our company. Then tell me if you see any gaps I should fill in.

Answer honestly. If your budget is two hundred dollars a month and four hours a week, say that. A strategy built on a fictional budget wastes everyone's time. If you have no competitors you know of, say that too — your VP will help you identify them.

The Strategy Brief

You are going to ask your VP to produce a marketing strategy brief — the kind of document a real marketing executive would present in their first month.

The difference between a useful strategy request and a useless one comes down to structure. Vague asks get vague answers.

Bad: "Make me a marketing plan." That gives your VP nothing to work with — no format, no time horizon, no decisions to make. You will get a generic ten-page document that could apply to any business.

Also bad: "Give me a comprehensive marketing strategy covering all channels, audiences, messaging, branding, competitive positioning, and quarterly goals." That is not structure — it is a wish list. When everything is a priority, nothing is.

Good: a request that specifies the deliverable, names the exact components you need, and states the constraints up front. Paste this:

Using everything you know about our company, competitors, channels, budget, and timeline, produce a marketing strategy brief with these four sections:

AUDIENCE SEGMENTS
Define 2-3 distinct audience segments we should target. For each one, give me:
- A specific description (not just "small businesses" -- tell me the job title, company size, industry, and what problem they are trying to solve)
- Why this segment is a priority for us right now
- How large this segment is relative to the others

MESSAGING PILLARS
For each audience segment, define 3-5 messaging pillars -- the core themes our marketing should hit. Each pillar needs:
- A one-sentence summary
- Why this message resonates with this specific segment
- One example of how it would show up in practice (a headline, a social post, an email subject line)

CHANNEL STRATEGY
For each audience segment, recommend which marketing channels to use. For each channel:
- Why this channel reaches this segment
- What kind of content works on this channel
- How often we should post or publish, given our budget and time constraints
- What success looks like in 90 days (specific metrics, not "increased engagement")

90-DAY ACTION PLAN
Give me a prioritized list of actions for the next 90 days. For each action:
- What it is
- Which segment and channel it serves
- How long it will take
- What I need to do vs. what you can help me produce

Be realistic about our budget and time constraints. I would rather have a plan I can actually execute than an ambitious one I will abandon in week three.

Each section exists for a reason. Audience segments force specificity — you cannot market to everyone effectively, and your VP needs to commit to who matters most right now. Messaging pillars create consistency. Channel strategy prevents the spray-and-pray approach. The 90-day plan turns strategy into action with deadlines you can actually hit. The final line — "be realistic about our budget and time constraints" — is the most important sentence in the entire prompt. Without it, you get recommendations built for a team of five with a ten-thousand-dollar monthly budget.

Review the Deliverable

Your VP will produce a brief. It will look polished and thorough. Read it like a skeptical business owner, not an impressed first-time user.

Start with the audience segments. "Small business owners" describes half the economy. "B2B SaaS founders with 10-50 employees who have outgrown word-of-mouth referrals and need a repeatable pipeline" is a segment — you can find those people, write for them, and measure whether your marketing reaches them. If the segments came back broad, push back.

Check the messaging pillars. Each segment should have distinct messaging. If all three segments get the same five pillars, the segmentation is not doing its job. Look at the channel strategy with your budget in mind — two channels executed well outperform five done poorly. Finally, check the 90-day plan for specificity. "Build social media presence" is not an action item. "Publish two LinkedIn posts per week targeting the SaaS founder segment using the 'operational efficiency' messaging pillar" is an action item. You can do it on Tuesday; you can check whether you did it on Wednesday.

Give your feedback directly, the same way you would talk to a real executive who brought you a first draft.

Good start, but I need changes:

1. The "growing businesses" segment is too broad. Split it into two: service businesses with 5-20 employees, and SaaS companies with 10-50 employees. They have completely different buying patterns.

2. The channel strategy has too many channels. I can commit to LinkedIn and email. Drop the rest for now -- we can add channels in Q3 if these two are working.

3. The 90-day plan needs specific weekly milestones. Break it into Month 1, Month 2, Month 3 with concrete deliverables for each.

Most strategy briefs take two or three rounds before they are solid. Three rounds is not failure — it is how you would work with a human VP too. And because of Memory (next section), those refinements carry forward. The VP will not make the same assumptions twice.

Meet Memory

Every decision you just made — the audience segments, the messaging pillars, the channels, the 90-day plan — your VP remembers all of it.

Cowork's Memory feature works in the background. As you and your VP discuss strategy, make decisions, and refine documents, Cowork captures the important details automatically. You do not need to save anything manually. You do not need to copy key decisions into a separate document. The VP retains what matters and references it in future conversations.

Memory is what makes your VP feel like a real employee instead of a stranger who keeps asking the same questions. It is a persistent context store scoped to a single Cowork Project, and both sides of the desk can write to it. The AI adds entries on its own when a conversation surfaces something worth keeping — a new audience segment, a constraint you mentioned in passing, a decision you made three tabs ago. You can open the Memory panel in the Cowork sidebar and add, edit, or remove entries by hand. Think of it as the running notebook your VP keeps on their desk, with your edit rights.

CLAUDE.md is the offer letter and the company handbook — stable instructions you author, review, and version yourself. Memory is everything your VP has picked up since the first day on the job — observations, preferences, decisions, the things a good employee just remembers without being told twice. One is deliberate and human-authored; the other is learned and AI-curated. CLAUDE.md sets the rules; Memory fills in the lived context.

Put Something Real Into Memory

You already have marketing material — a website, maybe a brand guide, maybe a folder of campaigns that worked and some that did not. Every one of those artifacts contains facts your VP should know, and right now none of them are in Memory.

The process: you hand Cowork your existing sources — website URLs, a brand guide PDF, samples of past emails, social posts, or ads — and your VP reads through all of it, sorts every fact into one of three buckets, and hands you back a classification report before anything is committed to Memory. You review the report, fix what the VP got wrong, and only then does the VP write the approved entries. Same way you would onboard a human hire, except the reading happens in minutes instead of a week.

The three buckets: BRAND (voice, tone, visual identity, forbidden words, positioning, tagline, company story), PRODUCT (what you sell, pricing, core benefits, differentiators, competitor landscape), and PAST-CAMPAIGN-HISTORY (what worked, what flopped, what audiences responded to, which channels you have tried, recent launches). Anything that does not fit those three does not belong in Memory yet.

One more thing: your sources will contradict each other. The brand guide says one tagline, the website header says another. A 2024 campaign used a voice the brand guide explicitly forbids. That is normal, and your VP should surface every contradiction rather than pick a winner on its own. You resolve conflicts, not the VP.

Copy the following prompt and paste it into your Cowork project:

I am going to share our existing marketing material with you so you can build a working picture of our brand, our product, and our marketing history.

Here is what I am giving you:
- Website URL(s): [paste one or more URLs here]
- Brand guide: [attach the PDF, or paste "none" if you do not have one]
- Past campaigns: [attach or paste samples -- emails, social posts, ads, landing pages, press releases -- whatever you have, 3 to 10 items is plenty]

Read every source end to end. As you read, classify each fact you find into one of exactly three Memory categories:

1. BRAND -- voice, tone, visual identity, forbidden words, positioning, tagline, company story.
2. PRODUCT -- what we sell, pricing, core benefits, differentiators, competitor landscape.
3. PAST-CAMPAIGN-HISTORY -- what worked, what flopped, which audiences responded to what, channels we have tried, recent launches.

If a fact does not clearly belong in one of those three, set it aside in a fourth bucket labeled UNCLASSIFIED and tell me why it did not fit. Do not invent a fourth Memory category.

Work through the following four phases in order.

PHASE 1 -- ANNOUNCE FILES
Before you read anything, tell me the Memory entries you intend to write once I approve.
Name each entry by category label and short title, for example "BRAND: voice adjectives."

PHASE 2 -- PAUSE FOR WRITES
Produce the full classification report in the conversation with these sections:
  BRAND (count: N) -- each fact on its own line, source in parentheses.
  PRODUCT (count: N) -- same format.
  PAST-CAMPAIGN-HISTORY (count: N) -- same format.
  UNCLASSIFIED (count: N) -- each item with a one-sentence note on why it did not fit.
  CONTRADICTIONS (count: N) -- every place two sources disagree. Name both sources; do not pick a winner.
  GAPS (count: N) -- categories where the sources gave you very little.
Show the full report and stop for my review.

PHASE 3 -- COMMIT ON APPROVAL
Do not write a Memory entry until I reply with a clear yes.
If I correct the classification, revise the report, show it again, wait for another yes.
Leave UNCLASSIFIED, CONTRADICTIONS, and GAPS out of Memory -- those stay in the chat for me to resolve.

PHASE 4 -- CLOSE WITH LOCATION
Once you save the approved entries, confirm each write on one short line.
Use this exact phrasing for every entry: "here is where it lives: Cowork Memory sidebar, entry titled [CATEGORY: short title]."
If a write failed or you saved under a different title, tell me that instead of guessing.

After PHASE 4 confirms the writes, your Project surfaces look like this:

your-cowork-project/
├── CLAUDE.md
└── (memory -- managed by Cowork, surfaced in UI)

Expect the classification report before anything lands in Memory. Once you have read it, corrected the misclassifications, resolved the contradictions, and told your VP which entries to commit, Memory gets its first real deposit — and every task from that point forward builds on material you verified yourself.

Here is Memory in action. Today, you tell your VP that your audience segments are "SaaS founders with 10-50 employees" and "service business owners with 5-20 staff." Next week, you open Cowork and say: "Draft a LinkedIn post promoting our new onboarding feature." You do not re-explain who the audience is. The VP already knows. It writes the post targeting SaaS founders, uses the messaging pillars you approved, keeps the tone consistent with the brand voice from your CLAUDE.md. That is Memory working across every conversation in this project. Memory also means your VP improves over time — the VP you have in three months will be meaningfully sharper than the one today, not because the technology changed but because it accumulated real knowledge about your business.

You're Still the Boss

Before you start handing your VP more assignments or building marketing tools, there is a mindset to establish. You are accountable for everything your VP produces.

When your VP drafts a blog post and you publish it, your name is on it. When your VP writes an email campaign and you hit send, your customers see your brand, not a disclaimer about AI. When your VP produces ad copy with a claim about your product, you are the one making that claim to regulators, competitors, and the public.

This is not hypothetical risk. It is how every business already works. If you hire a human VP and they put out a press release with a factual error, the correction comes from your company, not from the VP's personal account. The VP reports to you. You review the work before it ships. That does not change because the VP is an AI. The difference is that a human VP has instincts built from years of professional consequences — they know that claiming "our product cures X" without evidence invites lawsuits. Your AI VP has not been burned. It has no career to protect. That makes your review process more important, not less.

The Review Workflow

Every piece your VP produces travels the same path: Draft → Review → Revise → Approve → Publish.

• Draft — Your VP generates content from your brief, CLAUDE.md, and task-specific instructions.
• Review — You read the output against specific criteria. Not "does this feel right" but "does this meet my standards on each of these dimensions."
• Revise — Specific feedback. "The second paragraph feels off" is less useful than "the second paragraph claims we reduce costs by 40%, and I do not have data to support that number."
• Approve — You are comfortable putting your name on it.
• Publish — It goes live. You own it from this moment forward.

Two passes is the rhythm — five minutes for a social post, ten for a blog post, fifteen for something longer. The first few reviews take longer because you are calibrating. After two weeks, you will know your VP nails tone but overpromises on timelines, or writes great headlines but buries the CTA. A ten-minute review becomes a three-minute scan of the known weak spots.

When AI Gets It Wrong

Your VP produces excellent work most of the time. But "most of the time" is not a publishing standard. Know the failure modes so you can catch them during review.

• Hallucinated statistics. "Studies show that 73% of consumers prefer..." with no study behind it. The most common and most dangerous failure mode. If your VP cites a number, ask for the source. If it cannot produce one, cut the number.
• Made-up references. Companies, case studies, articles, or research that do not exist. A "2024 Forrester report on SMB marketing trends" that sounds perfectly real but was never published. Always verify external references.
• Brand-damaging tone. Technically fine but tonally wrong. A funeral home does not need "punchy" social copy. A children's education company does not need edgy humor.
• Legal claims you cannot support. "The best product in its category." "Guaranteed results." "Clinically proven." These are legal claims, not marketing flourishes — each requires specific evidence.
• Factual errors about your own products. If your product has changed since you wrote the CLAUDE.md, your VP does not know. Keep your brief current.
• Outdated best practices. "General" marketing knowledge means tactics that worked three years ago and are less effective today. Domain-specific context is your responsibility.

None of these failures mean your VP is bad at its job. They mean your VP needs a manager. That is you.

When NOT to Trust AI

Some content should never go through the draft-review-publish pipeline at all. Human-written, human-reviewed, human-approved — no AI in the drafting step.

• Legal statements. Terms of service, privacy policies, warranty language, regulatory disclosures. Your VP can help you brainstorm what to cover; the language must come from you or your attorney. A single wrong word in a warranty disclaimer can create liability your business insurance does not cover.
• Financial projections. Revenue forecasts, ROI claims, pricing commitments. Your VP will happily project 30% year-over-year growth based on nothing but pattern matching. Financial claims need real numbers from your real business.
• Competitor comparisons. "Our product is faster than CompetitorX" is a factual claim. If it is wrong, it is potentially defamatory. Comparative marketing requires careful framing your VP cannot validate.
• Crisis communications. When something goes wrong — a product recall, a data breach, a public complaint — your response shapes your brand for years. This is a judgment call about tone, timing, and transparency that requires human empathy.
• Anything involving personal data. Customer testimonials, case studies with named individuals, marketing that references specific people. Privacy regulations vary by jurisdiction and your VP does not know which ones apply to you.
• Final pricing decisions. Your VP can draft pricing page copy. It should not decide what the prices are.

The boundary is delegation versus abdication. A good CEO delegates the marketing plan. A reckless CEO abdicates responsibility for it. Delegation means you reviewed it. Abdication means you did not.

Your Review Checklist

Frameworks are abstract. Checklists are actionable. Here is the starting template. Every item maps to a real failure mode from the section above.

• Factual claims verified — Every statistic, percentage, and data point is traceable to a real source.
• Brand voice consistent — The content sounds like your company, not like a generic AI output.
• Legal and compliance reviewed — No unsupported claims, no guarantees you cannot back, no regulatory language.
• Call-to-action appropriate — The CTA matches the content and the audience's stage in the buying process.
• No hallucinated sources — Every referenced study, article, company, or quote actually exists.
• Audience targeting correct — The content speaks to the right audience, not a generic "business professional."
• Product details accurate — Features, pricing, availability, and capabilities match your current reality.
• Spelling and grammar clean — Read it out loud. If something sounds off, fix it.

Print it. Pin it next to your monitor. Use it every time you review a piece of content. Your checklist should evolve — the first time you catch a failure mode the list did not cover, add a new item.

In Part 2, you will encode these same standards as Rules your VP follows on every task, so the checklist runs before you even see a draft. Those Rules files will land in the .claude/rules/ directory inside your Cowork Project folder — the first on-disk subdirectory your Project grows. Accountability is the mindset; Rules are how the mindset gets enforced.

The Three Zones

Your checklist tells you what to check. This framework tells you when to check. Internal brainstorming is low-stakes. A press release is high-stakes. Treating them the same wastes your time on one end and risks your reputation on the other.

Zone 1: Automate Freely. Internal outputs that never reach a customer — brainstorming, research summaries, competitor notes, draft outlines, meeting prep, first drafts that will go through full review before publishing. Your VP can produce these without review. You read them when you use them.

Zone 2: Review Before Publishing. Outputs that reach your audience — blog posts, email campaigns, social posts, ad copy, website copy, landing pages, product descriptions, customer-facing presentations. Every item in this zone goes through your full review checklist before it goes live. No exceptions. "I am in a hurry" is not an exception. The checklist takes five minutes; repairing brand damage takes months.

Zone 3: Always Do Yourself. Outputs that carry legal, financial, or reputational risk no checklist can mitigate — legal statements, terms of service, financial projections, crisis communications, contracts, regulatory filings, statements involving named individuals, final pricing. Your VP does not touch these. You write them yourself or hire a professional. CEO decisions, not VP decisions.

What just changed

You wrote your first ./CLAUDE.md and approved the save into your Cowork Project root. You ran the interview-and-classify flow and approved a first deposit of BRAND / PRODUCT / PAST-CAMPAIGN-HISTORY entries into Cowork Memory. You produced a reviewed-and-revised strategy brief with audience segments, messaging pillars, channel strategy, and a 90-day plan — and Memory retained the decisions. And you internalized the accountability framework: a five-step review workflow, eight-item checklist, six failure modes to catch, six Zone-3 items to never delegate, and the three-zone scrutiny map.

Your Project folder now surfaces:

your-cowork-project/
├── CLAUDE.md
└── (memory -- managed by Cowork, surfaced in UI)

CLAUDE.md on disk. Memory in the sidebar. No .claude/ folder yet — that arrives in Part 2.

What is Next

Your VP knows the basics, carries the strategy in Memory, and you know how to keep it honest. What you do not have yet is a way to make your standards mechanical — to put "sound like us" and "run the review checklist before showing me a draft" into a file the VP reads on every task, without you having to type it in the prompt.

That is Part 2 — The Playbook. You will meet Rules — the first files to live in .claude/rules/ inside your Project — and codify your brand voice, content standards, process, and approval criteria as instructions your VP follows on every task. The review process gets easier when the first drafts are already in your voice.

Part 1 of 7 — Your AI VP of Marketing series. Next: Part 2 — The Playbook: Rules and Standards