Red teaming changed how I use AI

It’s great to use AI to build ideas. I also use it to break them.

Before any pitch deck or care program goes out, I run it through a red team prompt — skeptical investor, CIO, epic analyst, front-line coordinator, whoever would be hardest to convince. It finds the holes that an enthusiastic room never will.

I’ve taken it further. I’ll drop a meeting transcript in as context and red team the conversation itself. Where did we assume without verifying? Where did optimism outrun evidence?

AI has no ego in the outcome. No reason to stay positive in front of the team. That’s exactly what makes it useful here.

A yes-machine is a toy. A thought partner that keeps you honest is strategic infrastructure.

Try it yourself:

PROMPT

You are a skeptical

[investor / CMO / clinical informaticist

/ choose your critic].

You have seen this type of idea fail

before. Your job is to pressure-test what I'm about to share.

Deliver your critique in this structure:

1. Top 3 assumptions I haven't proven

2. The single most likely reason this fails

3. What a well-informed opponent would say to kill this

4. One blind spot I clearly haven't considered

Structure your output in a strategic incisive briefing with actionable recommendations to move forward.

Be direct. Do not soften. Do not offer encouragement.

[Paste your pitch, proposal, or meeting transcript]

Paste that before any pitch, proposal, or alongside a high stakes meeting transcript. See what comes back.