π― Expert Mode: The "Messaging Murder Board"
Scenario: You have a statement, a pitch, a press release line, or an executive talking point about to go external. You think it is ready. Before it leaves the building, you want five tough critics to tell you what breaks. You have 15 minutes.
Instructions:
1οΈβ£ Load the draft. Open your AI tool, paste in the message you want to test, and add any context you have (client, audience, sensitive history, prior public statements). Then set up the panel:
"You are a hostile but fair messaging murder board. I am going to give you a draft message that is about to go external. Convene five critics: a Skeptical Reporter, a Board Member or Funder, a Community Leader with lived experience of the issue, a Cautious Client Approver (CMO or General Counsel), and an Opposition Researcher. Here is the draft: [paste]. Here is the context: [paste, or write 'none provided']."
2οΈβ£ Run the panel. Ask each persona to stay in character and give their top two concerns specific to this draft, the one question they would ask in the room, and a one-line verdict.
"Run all five now. Keep them distinct. If two of them make the same point, push them back to their own vantage point. No generic feedback. Every concern must be specific to my draft. End each one with a verdict: would say it, would soften it, or would not let it out."
3οΈβ£ Force the synthesis.
"Now drop character. Give me the top 3 risks ranked by how much damage they would do, not how likely they are. For each risk, rewrite the line. Do not just flag it, fix it."
4οΈβ£ Find the landmine.
"Flag any single line that could become the story, get weaponized, or breach trust. Quote the exact line and tell me why. If there is nothing, say so plainly."
5οΈβ£ Protect the strongest part.
"Last thing. Tell me the single strongest element in the draft so I keep it when I revise."
Suggested Tool: Claude or ChatGPT
Why You'll Love It
- It finds the weak line before a reporter or a screenshot does. The whole point is to fail in private, where a fix costs you five minutes instead of a news cycle.
- It replaces "looks good to me" with five specific objections. Most internal reviews are too polite to catch the thing that actually breaks. This panel is built to be impolite.
- It rewrites, it does not just critique. You walk away with fixed lines you can paste back in, not a list of vague worries.
The best messaging does not come from the first draft. It comes from the draft that already survived the room. This prompt hands you the room before you need it.
Talk soon, Chris
Want to take this further? Here is the Cowork version.
The Murder Board pressure-tests your message. The Cowork version pressure-tests your decision. It convenes a five-seat Board of Directors that reads your actual context, argues both sides of a real call, and drops a decision memo straight into your folder.
Setup (one time):
- Pick the real decision you are wrestling with. Phrase it as a yes/no or an either/or, not a vague topic.
- Gather the context that matters (relevant files, notes, prior plans, numbers) and point Claude at the folder.
- Decide what winning looks like and what you are leaning toward, so the board has something to push against.
- Optional: name the folder where the memo should be saved (default: your KNOWLEDGE folder).
Run this Cowork command:
"Convene my Board of Directors on this decision: [state it as a yes/no or either/or]. Here is what I am weighing: [the options, including doing nothing]. Winning looks like [outcome]. I am leaning toward [gut call]. Do the following:
Read the context I point you to in [folder or files] so the board argues from my real situation, not generic advice.Convene five distinct seats: the Operator (Chair), the Brand and Narrative Strategist, the Creator-Economy Operator, the Audience Advocate, and the Contrarian whose job is to kill the idea.Have each seat give a sharp read, the one question they would put to me, and a call (Go, Go but, Not yet, or Don't). Keep them genuinely distinct.Have the Chair synthesize: where the board agrees, where it splits, a clear recommendation, and the strongest dissent preserved right next to it.Save the result as a decision memo, titled by the decision and dated, in [target folder]."
What you'll get:
- Five distinct points of view on your decision, each argued from its own mandate, not a popularity vote
- The real fault line named honestly, so you see the disagreement instead of a smoothed-over average
- A clear recommendation with the strongest counter-argument preserved beside it, so you know what you would be betting against
- The single highest-leverage next step to take, or the one fact you still need to go find
- A clean decision memo saved to your folder that you can revisit when the call gets second-guessed
Want it on autopilot? Say "turn my Board of Directors into a saved skill" and Claude will set it up so you can convene the board on any decision with one line, any time.
Here are a few articles from the past week to help keep you up to speed:
1. Axios AI+NY Summit: Brands Get the Message, AI Is Rewriting All the Rules Axios June 8, 2026
Comms leaders from JPMorgan and PSEG told the room that a brand's reputation now sits in the hands of AI bots that can grade, alter, and distribute information about it with no company input. One team has started writing two versions of every story, one for humans and a separate bot-oriented piece for the machines. The takeaway for comms pros is blunt: what you produce matters more than ever, because the machine reading it never forgets and never asks a follow-up.
2. Everything-PR Publishes First AI Communications 100β
O'Dwyer's PR News June 12, 2026
A new list maps the 100 people whose decisions now influence whether your work gets surfaced when AI engines answer questions about your category. The framing is the part worth sitting with: a placement in Wired is now read by the editor and a model's training pipeline at the same time, and a crisis statement in the Times lives on inside AI summaries of your reputation for the length of the training window. Visibility is no longer just about the human reader.
3. What to Know About PR 3.0 and the Intelligence Eraβ
PRSA June 2026
PRSA argues there is now an intelligence layer sitting between every communicator and every audience, synthesizing what the world believes about your organization and serving it up to people who never visit your site. It names the new craft skills directly: dual-purpose drafting, entity-rich language, structured data, and consistency audits across your whole ecosystem. If you want a vocabulary for explaining to leadership why comms work is changing, this is a clean one.
4. Press Release Format Breakdown: What Journalists Want in 2026 Agility PR Solutions (via Daily PR Brief) June 15, 2026
A practical breakdown of how to structure a release for two readers at once: a journalist skimming in seconds and an AI pulling context for an answer it will generate later. It pairs directly with this week's theme, because the same line has to work for a human in a hurry and a model that will quote you out of context. Most releases are still written for neither.
5. The Deepfake Era Has Arrived and PR Is the Front Lineβ
O'Dwyer's PR News Reported May and June 2026
Deepfake fraud has spiked roughly 3,000%, a single viral hoax can knock 16% off a brand's reputation, and yet a majority of companies still have no formal plan to respond. The uncomfortable part for comms teams: even after a fake is debunked, the smear it carried often travels further than the correction. This is exactly the kind of scenario you want to pressure-test before it is live, not during.
β