What a Decision Rehearsal Is
A decision rehearsal is a structured process that runs a high-stakes decision through systematic analysis before the organization commits to it. Unlike brainstorming sessions, strategy offsites, or unstructured AI conversations, a rehearsal follows a governed pipeline where every stage has defined inputs, required outputs, and validation gates.
The concept is simple: if a decision is worth committing your organization to, it is worth rehearsing first. Rehearsal doesn't mean hesitation. It means disciplined preparation — the kind that turns "I think this will work" into "here's the evidence, here's what breaks, here's our countermeasure for each risk, and here's our verdict."
A rehearsal takes 20-30 minutes. A wrong decision takes months to unwind — if it can be unwound at all.
Why Pre-Mortems and Scenario Planning Aren't Enough
Pre-mortems are valuable. You imagine the decision has failed and work backward to find the reasons. But pre-mortems are constrained by the imagination of the people in the room. They surface the risks people can think of — not the cascading failures hidden in dependency structures nobody mapped.
Scenario planning has the same limitation. You define three or four scenarios and analyze each one. But those scenarios are selected by humans with bounded rationality and cognitive biases. The scenario that actually materializes is often the one nobody wrote down — because it emerges from the interaction of dependencies that weren't visible at the planning stage.
The difference: A MAIA rehearsal doesn't start with imagined failures. It starts with dependency mapping. It identifies what the decision actually touches, then generates stress-test scenarios from the structure itself. The risks aren't imagined — they're derived from the system.
The Rehearsal Model: Structured, Governed, Evidence-Scored
A MAIA rehearsal operates on three principles that separate it from unstructured analysis:
Structured
Nine stages, executed in order. Each stage has a defined purpose, required artifacts, and a validation gate. You can't skip to "what should we do?" without first mapping what the decision touches and how disruptions propagate through it. The structure prevents the most common failure mode of strategic analysis: jumping to solutions before understanding the system.
Governed
Every stage output is hashed and locked to memory. Every agent operates within enforced role boundaries — scope constraints, forbidden domains, and output requirements. A validation engine (RRV) checks 32 rules at every stage transition. External evidence requires your explicit permission before any search, and your approval before any finding is incorporated. Governance isn't optional. It's architectural.
Evidence-Scored
Every claim in the final decision brief traces back to evidence. Source admissibility rules determine what counts — regulatory filings, central bank publications, audited financials, not opinion pieces or blog posts. Confidence bands are assigned. Corroboration is required. The result is a verdict you can defend to a board, not a conversation you have to summarize from memory.
How a MAIA Rehearsal Works
MAIA delivers a rehearsal in two passes within a single session:
Pass 1: The Prose Rehearsal
You describe your decision in a conversation with Claude. MAIA's pipeline activates — 45 specialized agents work through nine stages of analysis. You guide the process: approving evidence, selecting which disruptions to simulate, shaping the analysis at each decision point.
This pass is a dialogue. You bring the context. MAIA brings the structure. Stage by stage, the analysis deepens — dependencies mapped, timeline established, structural fragility exposed, cascades simulated, countermeasures designed, verdict rendered.
Pass 2: The Visual Package
When the rehearsal is complete, MAIA generates a Visual Package — a single HTML file with eight interactive cards built from the analysis you just shaped. Download it. Open it in any browser. Share it with your board. No account required. No software to install. It works offline.
Eight cards: Command Center, Data Scope, Evidence, Analysis Pipeline, Cascade Propagation, Stabilizers + Readiness, Decision Brief, and Export with full provenance.
What You Get
A completed rehearsal produces structured, referenceable deliverables:
- Dependency map — Eight categories of dependencies with entity universe, pathway depths (1/3/5-step), and critical path identification
- Cascade simulation — How selected disruptions propagate through your dependency structure, step by step, with cross-scenario pattern detection
- Countermeasures — Stabilizers designed for specific cascade paths, scored by feasibility, with residual risks disclosed
- Decision brief — Viability verdict with optimal path, resilience score, risk register ranked by severity, and monitoring indicators with trigger thresholds
- Visual Package — A single downloadable HTML file with eight interactive cards covering the full analysis. Portable, shareable, works anywhere.