I asked Claude AI to make me as much money as possible

Overview: Why “yes-man” AI fails (and what to do instead)

The video critiques using “yes-man” AI (e.g., a standard Claude prompt) for business decisions. It argues that you often get a mirror of your own framing—leading to the wrong product, wrong pricing, or wrong messaging.

Instead, it promotes a structured “pressure test” workflow called Claude Counsel.

Core framework: Claude Counsel (advisor board + console verdict)

Claude Counsel works like a multi-agent decision process:

• 5 AI advisors (different attack angles)

  • Contrarian: find fatal flaws first (what will break and why)
  • First-principles: clarify the real problem being solved (what you’re actually trying to accomplish)
  • Expansionist: find overlooked upside/opportunity
  • Outsider: read the situation with zero context to spot obvious gaps you missed
  • Executor: convert strategy into the next concrete action (today’s step)

• 5 peer reviewers

  • Review all advisor outputs without knowing who said what
  • Evaluate quality/validity of reasoning rather than author credibility

• 1 final “console verdict”

  • Produces:
    • One verdict
    • One clear next step
  • Avoids “balanced pros/cons” that require interpretation

Outcome: more reliable recommendations than a single AI response because no one perspective dominates.

Examples: actionable business decisions

Example 1: Product idea validation with a hard KPI/time constraint

Scenario: Build an email content planner (interactive mini web app) and launch for $47 after ~3 weeks.

Main lessons:

• AI may validate automatically (“yes”), missing common execution killers:
  • unclear positioning
  • target audience too broad
  • pricing too ambitious without an existing audience/email list
• Adding a time-bound goal clarifies execution reality:
  • “Earn $1,000 in the next 30 days” with no social media following

Console-guided findings:

• A 30-day window with no audience is a tight constraint:
  • risk of building weeks for a product nobody wants
  • then spending time finding buyers
  • likely underperformance by day 30 (illustrated risk: only ~$200 sales before giving up)
• The framing shifts the problem from “product” to distribution.

Recommended operating plan (parallel execution + warm outreach):

• Track A: distribution (2 hours/day)
  • engage on Threads
  • grow presence
  • reach out via warm channels (friends, communities, email list, Discord, Twitter/Reddit/LinkedIn groups)
• Track B: validation + selling
  • compile 50–100 names of potential users
  • send a simple outreach message asking if they want to learn more
  • position it as the easiest path to the first $1,000 in 30 days

Execution KPI gate: If you can’t name 50 people, the timeline to $1,000 will likely slip (“deadline gets longer”).

Emphasis: the workflow prioritizes concrete “next-step” outputs over generic encouragement.

---

Example 2: Pricing decision (conversion-risk framing + offer packaging)

Scenario: Mini web app priced at $27; suggestion to raise to $97.

Console logic:

• Contrarian: raising price without changing the offer is a gamble; if conversion drops and traffic is low, you lose revenue and can’t correct quickly.
• First-principles: if the buyer saves time, the high price can be justified (example: saving ~3 hours/month of manual work).
• Expansionist: instead of only increasing the base price, create a second tier/bundle:
  • keep $27
  • add $97 as a “done-for-you” setup/templates bundle
• Outsider: a sales page may look like a feature list instead of showing the time-saved math.
• Executor: don’t jump straight to $97 without evidence:
  • update positioning to lead with time savings
  • add a result-focused testimonial
  • test a $47 step as an intermediate pricing test
  • revisit after 30 days

Timeline/KPI: reassess in 30 days; use staged pricing tests rather than one big jump.

---

Example 3: Pre-launch failure-mode analysis

Before launching any offer/content, ask:

• “What is the most likely reason this will fail in the next 30 days?”

It’s intentionally not “what could go wrong.” The goal is to surface the highest-probability blocker, often positioning or audience-market fit.

The console credits its contrarian/outsider roles with finding issues the builder may miss.

Operating recommendations (process playbook)

The video recommends running a “console verdict” before committing to key decisions:

Use Claude Counsel for:

• “Is my idea worth building or is there a faster version?”
• “Is my pricing right?”
• “Is my sales page clear for someone who’s never heard of me?”
• “What is the most likely reason my first launch fails?”
• “Should I expand the product line or go deeper?”
• “Am I solving the right (most obvious) problem?”
• “What should I do next today?”

Input quality rule

Provide context for better output:

• what the product is
• who it’s for
• price
• marketing plan
• biggest concern / what you’re worried about
• include your draft page when asking for sales-page critiques

Implementation (how the workflow is deployed operationally)

High-level steps to use Claude Counsel:

1. Recommend Claude Pro.
2. Create a new Claude project (instructions + files persist).
3. Upload the shared file and paste the system prompt into project instructions.
4. Start a new chat in that project and append: “ask the council”.
5. Expect response time: ~60–90 seconds depending on complexity.

Key KPIs / targets explicitly mentioned

• Revenue target: $1,000 in 30 days
• Pricing points to test: $27 → $47 → $97
• Sales page positioning concept: show time-saved math
• Outreach quantity: 50–100 names (warm-list baseline)
• Activity requirement: 2 hours/day engaging/growing on Threads
• Decision timeline: revisit pricing/strategy in 30 days
• Illustrated risk metric: possibly only ~$200 sales by day 30 if distribution is missing

Presenters / sources

• Presenter in the video: Not explicitly named in the subtitles (referred to themselves as the channel host).
• Creator/source of Claude Counsel: Ollie Lehman
• AI system referenced: Claude (Claude AI / Claude code) and Claude Cowork