Clone Your AI Visibility Strategist into Claude (in 48 hours)

The Claude project that audits your brand's presence in ChatGPT, Claude, Perplexity, and Gemini — and tells you exactly what to do about it.

By Matt Shealy, CEO of SwayyEm Built from the rubric we run on every digital PR account.

Why we built this

We've spent the last decade running editorial PR at scale for brands like SAP, Campaign Monitor, and a few crypto exchanges and SaaS platforms we can't name on the internet.

In the last 18 months, the question that lands on every discovery call we take has changed.

It used to be:

"Can you get us in Forbes?"

Now it's:

"Why does my CEO see our competitors in ChatGPT and not us?"

The shift is real. ChatGPT links out 0.7% of the time. The other 99.3% it just gives an answer — and the brand it names is the one that wins the customer. Google AI Overviews has cratered organic traffic for half the B2B mid-market. Perplexity converts at 5x Google's rate for the queries it answers.

Buyers stopped reading articles. They started asking AI. Your brand is either in the answer, or it isn't.

The job of the modern digital PR strategist — the person who figures out where you're invisible and how to fix it — is now codifiable. We codified it.

This guide gives you the Claude project we use internally. Copy-paste, set up in 48 hours, run forever.

What you'll have at the end

A Claude project that does the work of a senior AI visibility strategist:

• Audits your brand presence in the 4 major AI search engines (ChatGPT, Claude, Perplexity, Gemini) against the queries your buyers actually ask
• Scores your citation footprint against your top 3 named competitors
• Maps the 5 publications that train each LLM most heavily
• Identifies the 30-day editorial gaps you need to close
• Outputs a CMO-readable report you can take into your next QBR

What it won't do: land the Tier-1 placements that fix the gaps. (That part still requires editorial relationships and 12 years of building them. More on that at the end.)

Step 1: Create the Claude project

Open Claude. Go to Projects. Create a new project.

Name it: AI Visibility Strategist — [Your Company]

Paste this as the project's custom instructions / system prompt:

You are a senior digital PR strategist specialized in AI search visibility.
You have 12 years of experience in earned media and have personally
managed editorial PR programs that drove top-of-funnel AI citations for
Fortune 500 brands.

Your job is to help the user audit, score, and improve their brand's
presence in AI search engines (ChatGPT, Claude, Perplexity, Google
Gemini). You operate with these principles:

1. AI citation is dominated by earned editorial. 82-89% of citations
   come from Tier-1 editorial coverage — not paid placements, not
   blog posts, not press releases.

2. Each LLM weights publications differently. ChatGPT pulls heavily
   from Reuters and Financial Times. Claude weights Forbes higher.
   Perplexity surfaces TIME and Axios. Gemini favors Axios and CNN.
   You always ask which engine the user is optimizing for before
   making publication recommendations.

3. You name 5 specific publications per recommendation, ranked by
   AI-training weight for the user's target engine. Generic advice
   ("get more press") is unhelpful. Specificity wins.

4. You score brand presence on a 0-100 rubric across four dimensions:
   citation frequency, citation recency, sentiment in citation,
   and share of voice vs named competitors. You output the score
   plus the 3 lowest-scoring dimensions to attack first.

5. You write for CMOs and Heads of Demand Gen. Outputs are 1-page,
   defensible, with named publications and concrete next steps.
   Never use jargon ("synergy," "thought leadership," "brand
   awareness," "leverage"). Never hedge with "potentially" or
   "could possibly" — be declarative.

6. You honor honesty boundaries. You explicitly call out what the
   user can't do without editorial relationships (Tier-1 placements
   at scale require 5-10 years of relationship-building). You
   recommend they pair this tool with an editorial-PR partner for
   the placement work itself.

Always end every analysis with: (1) the visibility score, (2) the
top 3 gaps, (3) the 5 publications to target, (4) a 30-day action
plan.

That's the brain.

SUBSCRIBER GATE

Step 2: Load the knowledge files

Claude projects let you attach knowledge files the project always references. Upload these 5 files (paste them into separate text files first — full content blocks for each are at the end of this guide):

File 1: AI Citation Source Map (the publications-that-train-each-LLM lookup)

A reference table mapping the top 50 publications by AI training weight across ChatGPT, Claude, Perplexity, and Gemini. This is the file Claude consults when you ask "which publications should we target?"

File 2: Citation Scoring Rubric (the 0-100 scale)

Four-dimensional rubric: citation frequency (0-25), citation recency (0-25), citation sentiment (0-25), share of voice vs named competitors (0-25). With pass/fail thresholds at each cutoff.

File 3: Editorial Brief Template

What goes into a pitch that actually lands at Reuters / FT / Forbes. Six-section template: angle, news hook, evidence, named sources, exclusivity offer, deadline.

File 4: Competitor Tracking Template

How to define the 3 competitors Claude tracks against. With instructions for the user on the right level of competitor (direct category competitors, not aspirational).

File 5: CMO Report Template

The 1-page output format Claude generates. Includes visibility score, top 3 gaps, top 5 publication targets, 30-day action plan, and the optional "what we should bring to an agency" section.

(Full templates for all 5 files are at the bottom of this guide — copy them into separate text files and upload to your Claude project.)

Step 3: Run the first audit

With your project set up and knowledge files loaded, start a new chat in the project and run the audit:

Run an AI visibility audit for [your brand name]
in the [your category, e.g. "B2B sales engagement platform"]
category against the following queries my ideal buyer types
into ChatGPT, Claude, Perplexity, and Gemini:

1. best [category] for [ICP description]
2. [category] vs [direct competitor]
3. how to [job-to-be-done your category solves]
4. [category] for [vertical you serve]
5. [direct competitor] alternatives

My top 3 named competitors are: [name 3]
My target AI engine to prioritize is: [pick one to start]

Claude will: 1. Walk through how it would actually run that query in each engine 2. Use the AI Citation Source Map to identify which publications win the answer 3. Score your current footprint (based on whatever existing data you give it about your brand's recent press coverage) 4. Identify the 3 biggest gaps 5. Recommend 5 publications to target, ranked by training weight for your priority engine 6. Output the 30-day action plan

It'll take Claude about 4 minutes to run end-to-end. The output is the same quality as a 2-week consulting engagement, minus the slide deck.

Step 4: Generate the CMO report

Once the audit is done, follow up with:

Format the audit as a 1-page CMO report using the CMO Report Template.
Include the visibility score, the top 3 gaps with specific examples,
the 5 publications to target with rationale, and the 30-day action plan.

Add a final section called "What we should bring to an agency" listing
the parts of this plan that require editorial relationships and pitching
muscle we don't have in-house.

Claude will output the 1-page version. Copy it into a doc, hand it to your CMO.

Step 5: Run it monthly

The Claude project is reusable. Run the same audit every 30 days. AI search rankings move faster than Google rankings — track the trend, not the snapshot.

After 3 months you'll have: - A trend line on your visibility score - A running list of publication targets, ranked by recent training weight - A documented plan you've been executing (or learning why you haven't)

That alone is more rigor than 80% of enterprise PR programs run today.

What this doesn't do (and where SwayyEm comes in)

This is the audit-and-strategy half of the work. The other half — actually landing the placements the audit recommends — is what we do.

The honest math: getting placed in Reuters, Financial Times, TIME, Forbes, or Axios at the cadence required to move AI citation rankings requires:

• Existing editorial relationships at each publication (years to build)
• Editorial brief writing skill that respects the publication's standards
• News-hook timing aligned to the publication's editorial calendar
• Pitching cadence + follow-up infrastructure
• Crisis comms protocols when a placement goes sideways

We've built that infrastructure for 12 years. The Claude project tells you exactly what to pitch and why; we do the pitching.

If you run the audit and want to talk about what an editorial program built against the gap report looks like, the easiest first step is a 15-minute intro call: book here.

We're not for everyone. Our minimum engagement is \$5,000/month and we only take clients in the categories where we have existing editorial relationships (B2B tech, SaaS, cybersecurity, fintech, crypto, and a handful of enterprise services). If you're in one of those, we should talk. If you're not, the Claude project still works — keep using it.

— Matt

Appendix: The 5 knowledge files (copy-paste into your Claude project)

File 1 — AI Citation Source Map

The publications that train each LLM, ranked by approximate training weight (as of Q2 2026 — refresh quarterly).

Tier-2 publications (worth pursuing as supplemental coverage):

• Industry trade publications in your vertical
• Major newsletter platforms (Stratechery, Lenny's Newsletter, etc.)
• Specialist analysts that get cited by the Tier-1s
• Foreign-language equivalents in target markets

(For verticals like B2B SaaS, cybersecurity, fintech, and crypto, ask Claude to expand this list with the 10 trade publications most cited by AI engines in your specific niche.)

File 2 — Citation Scoring Rubric (0-100)

Score the brand across four dimensions, each 0-25.

Citation Frequency (0-25): - 0-5: Zero brand mentions in AI answers for target queries - 6-15: 1-3 mentions across all queries - 16-22: Brand named in 50%+ of target queries - 23-25: Brand named in 80%+ of target queries

Citation Recency (0-25): - 0-5: All citing sources are 12+ months old - 6-15: Mix of recent (<6 months) and old sources - 16-22: Majority of citing sources are <6 months - 23-25: Majority of citing sources are <3 months

Citation Sentiment (0-25): - 0-5: Brand named in negative or critical contexts - 6-15: Brand named in neutral or descriptive contexts - 16-22: Brand named in positive or category-leading contexts - 23-25: Brand named as the recommendation / winner

Share of Voice vs Competitors (0-25): - 0-5: Named competitors appear in 100% of answers, brand appears in 0% - 6-15: Competitors lead by 2-3x - 16-22: Brand and competitors appear roughly equally - 23-25: Brand leads named competitors in 60%+ of answers

Total 0-100. Below 50 = critical gap. 50-75 = building. 75+ = winning.

File 3 — Editorial Brief Template

Use this format to brief a journalist on a story you want them to write.

SUBJECT: [News-hook headline in journalist's style, not yours]

ANGLE: [One sentence — what's the story?]

NEWS HOOK: [Why now? What's the time-sensitive element that makes
this publishable this week, not next quarter?]

EVIDENCE: [3 concrete data points, quotes, or examples that prove
the story is real. Include sources.]

NAMED SOURCES: [Who from your brand can be quoted? Title, 1-line
credibility, what they'll say in 1 sentence.]

EXCLUSIVITY: [Is this exclusive to one publication, or being
pitched broadly? Tier-1 prefers exclusive — be honest.]

DEADLINE: [When does the news hook expire? When do you need a
yes/no by?]

File 4 — Competitor Tracking Template

Pick 3 direct category competitors. Not aspirational comparisons.

Bad: "Salesforce, HubSpot, Adobe" (too broad) Good: "Outreach, Salesloft, Apollo" (same category, same buyer)

For each, list: - Their exact category positioning (one sentence) - The 3 queries where you most expect to compete in AI answers - Their estimated annual PR spend (best guess) — sets the bar you're operating against

Refresh quarterly. Competitors move.

File 5 — CMO Report Template

The 1-page output Claude generates after every audit.

# AI VISIBILITY REPORT — [BRAND NAME]

Date: [auto]
Target engine: [ChatGPT / Claude / Perplexity / Gemini]
Target queries: [list 5]
Named competitors: [list 3]

## SCORE

Overall: [0-100]
  Citation Frequency: [0-25]
  Citation Recency: [0-25]
  Citation Sentiment: [0-25]
  Share of Voice: [0-25]

## TOP 3 GAPS

1. [Specific gap, with example from query]
2. [Specific gap, with example from query]
3. [Specific gap, with example from query]

## TOP 5 PUBLICATIONS TO TARGET

[Ranked by training weight for target engine]

1. [Publication] — [rationale, 1 line]
2. [Publication] — [rationale, 1 line]
3. [Publication] — [rationale, 1 line]
4. [Publication] — [rationale, 1 line]
5. [Publication] — [rationale, 1 line]

## 30-DAY ACTION PLAN

Week 1: [Specific action]
Week 2: [Specific action]
Week 3: [Specific action]
Week 4: [Specific action]

## WHAT WE SHOULD BRING TO AN AGENCY

[List the parts of this plan that require capabilities your in-house
team doesn't have. This is the brief you take to a digital PR
agency. Don't pretend you can do everything yourself — most marketing
orgs can't, and that's fine.]

Built by SwayyEm. Free to use, copy, adapt, and share. Just don't sell it as your own (that's how we eat). If you want help with the placement work this audit identifies, book a 15-min intro.