Course Objective & Scope
Both human teams and default AI assistants suffer from agreement bias: colleagues hold back out of politeness or hierarchy, while LLMs are trained to be agreeable (sycophancy). This course combines the strategic discipline of Red Teaming with structured LLM prompting so leaders can stress-test strategies, product plans and investment memos before they reach the boardroom or the market.
Participants run their own real proposals through adversarial AI protocols—Steelmanning, Competitor Response, Sceptical CFO and Pre-Mortem Personas—to expose blind spots and harden their plans.
Who Should Attend?
- Managers and leaders authoring strategy, product, investment or transformation proposals
- Project/programme managers, enterprise architects and internal consultants
- Decision-makers looking to use AI for deeper critical thinking rather than shallow drafting
- Access to any standard LLM assistant (ChatGPT, Claude or Copilot) is sufficient
Key Takeaways
- Override LLM sycophancy using blind-evaluation and adversarial system prompts
- Steelman opposing arguments to test their proposal against the strongest possible critique
- Run multi-persona Red Team simulations (competitor CEO, regulator, churned customer, CFO)
- Embed AI as a neutral 'devil's advocate' in team meetings to surface dissent without personal friction
- Turn Red Team findings into concrete plan revisions, risk mitigations and pre-emptive Q&A rebuttals
Curriculum
Strategic Red Teaming and Overcoming LLM Sycophancy
- The Red Team mindset: attacking a plan in private so it survives contact with reality
- Why LLMs default to flattery and how blind/adversarial prompting forces honest critique
Steelmanning and Automated Assumption Hunting
- Building the strongest possible case for the opposite decision (Steelmanning)
- Prompting AI to extract unstated assumptions, logical leaps and second-order consequences
Multi-Persona Adversarial Simulations and Competitive Wargaming
- Stress-testing proposals through 4 lenses: Aggressive Competitor, Sceptical CFO, Churned Customer and Frontline Ops
- Simulating competitor counter-moves and market friction
Using AI as the 'Zero-Ego Devil's Advocate' in Team Meetings
- Depersonalising critique by making a structured AI Red Team pass a standard pre-read ritual
- Filtering out noisy AI objections to focus on genuine structural vulnerabilities
Live Proposal Stress-Test and Red-Team Playbook Workshop
- Running a real participant strategy or project memo through a 5-stage AI Red Team protocol
- Leaving with a hardened decision memo and a reusable Red-Team Prompt Library
Other Courses in This Category
-
Defining the real root problem instead of chasing symptoms; covers the Problem Statement Canvas, MECE issue trees, 5 Whys, scope boundaries and aligning stakeholders before jumping to solutions.
- Live Course
- 5 modules
-
Hypothesis-driven analysis, argument mapping (claim-evidence-warrant), deductive and inductive reasoning, spotting logical fallacies and assumption mapping; participants solve business cases systematically.
- Live Course
- 5 modules
-
Kahneman's System 1/System 2 thinking, confirmation bias, sunk cost fallacy, anchoring, groupthink and decision noise; teams design practical debiasing checklists and decision hygiene routines.
- Live Course
- 5 modules
-
Testing the quality of evidence across reports, market studies, internal metrics and AI outputs; covers correlation vs causation, survivorship bias, base rates and source triangulation.
- Live Course
- 5 modules
-
Defining weighted decision criteria (MCDA) before evaluating options, separating knockout must-haves from scored trade-offs and building decision matrices on a real organisational choice.
- Live Course
- 5 modules
-
Overcoming brittle single-point forecasts with 2x2 scenario matrices, early-warning signposts, trigger-based contingency plans and Gary Klein's Pre-Mortem method on a real initiative.
- Live Course
- 5 modules
Get in touch
