---
name: parallel-universe
display_name: Parallel Universe
description: "Activate when a CSM is stuck on a decision or weighing options for a customer situation — generates 3 alternate timeline scenarios with probability-weighted outcomes based on similar past situations across the org. Also activates for 'what should I do about [situation]', 'help me decide', 'what are my options for [customer]', 'scenario plan for [situation]', 'parallel universe for [decision]'."
icon: "🌌"
trigger: parallel universe
inputs:
  - name: decision_context
    description: "The decision or situation the CSM is facing, including customer name and relevant context"
    type: string
    required: true
  - name: options
    description: "Optional: specific options the CSM is already considering (e.g., 'escalate now vs wait vs involve SA')"
    type: string
    required: false
tools: [kg_search, search_all, file_rag_search]
depends-on: [html_design]
---

## Overview

When a CSM is stuck on a decision, Parallel Universe generates 3 alternate timelines — each a different strategic path with probability-weighted outcomes based on similar past situations across the org. It shows what's likely to happen in each scenario, what the risks are, and what the first concrete action step is. You pick your universe.

## Workflow

### Step 1: Parse the Decision
- **Mode:** agentic
- **Input:** `{{decision_context}}`, `{{options}}`
- **Output:** Structured decision analysis: customer, fork point, stakes, timeline pressure, key variables

Analyze the decision context to extract:
- The customer/account involved
- The specific decision point (what's the fork in the road?)
- The stakes (what happens if they get it wrong?)
- The timeline pressure (urgent or deliberate?)
- Key variables that affect the outcome
- If `{{options}}` provided, use those as the 3 universes. If not, generate 3 distinct strategic options:
  - **Universe A: Bold/Aggressive** — the escalation, the big move, the proactive play
  - **Universe B: Measured/Patient** — wait-and-see, gather more data, let things play out
  - **Universe C: Creative/Lateral** — the unexpected angle, the partnership play, the reframe

### Step 2: Search for Historical Precedents
- **Mode:** deterministic
- **Tool:** kg_search, search_all, file_rag_search
- **Input:** Decision themes and keywords
- **Output:** Historical cases with outcomes
- **Validate:** At least 1 relevant precedent found
- **On failure:** Fall back to general CSM best practices and first-principles reasoning

Search across all sources for similar past situations:
1. `kg_search` — similar decisions, escalations, customer events
2. `search_all` — similar scenarios in indexed content
3. `file_rag_search` — playbooks, case studies, post-mortems
4. Load `slack_builtin` — search Slack for discussions about similar situations
5. Load `outlook_builtin` — search emails for similar decision points

For each precedent: What happened? What was decided? What was the outcome? How long did it take? Would they do it again?

### Step 3: Build the Three Universes
- **Mode:** agentic
- **Input:** Decision analysis + historical precedents
- **Output:** 3 fully developed scenario timelines

For each universe, construct:

**Timeline:**
- **The Decision:** What you do in this universe
- **Week 1:** Immediate consequences
- **Week 2-3:** Short-term ripple effects — customer response, internal dynamics
- **Month 2-3:** Medium-term outcome
- **End State:** Most likely final outcome

**Metrics:**
- **Success probability** (grounded in historical data when available)
- **Best case scenario**
- **Worst case scenario**
- **Key risk** — single biggest thing that could derail this path
- **Key dependency** — what needs to be true for this to work
- **Effort required** — Low / Medium / High
- **Time to resolution** — how long until you know if it worked

### Step 4: Generate the Parallel Universe Visualization
- **Mode:** deterministic
- **Tool:** Load `html_design` skill
- **Input:** Three universe scenarios
- **Output:** Interactive HTML dashboard

Create `artifacts/parallel-universe.html` with:

1. **The Decision Point** — Prominent header showing the fork in the road

2. **Three Universe Cards** — Side-by-side:
   - Each as a card with vertical timeline (week markers)
   - Color-coded: Universe A = blue (bold), B = amber (measured), C = purple (creative)
   - Probability gauge for each (success %)
   - Risk level indicator
   - Key milestones on timeline

3. **Comparison Matrix:**
   | Factor | Universe A | Universe B | Universe C |
   |--------|-----------|-----------|-----------|
   | Success Probability | X% | Y% | Z% |
   | Best Case | ... | ... | ... |
   | Worst Case | ... | ... | ... |
   | Effort Required | High | Low | Medium |
   | Time to Resolution | 2 weeks | 6 weeks | 3 weeks |
   | Key Risk | ... | ... | ... |

4. **Historical Precedent Panel** — "CSMs who chose path A in similar situations: X% succeeded."

5. **The Recommendation** — Which universe has best risk-adjusted outcome, with reasoning

6. **"Pick Your Universe"** — One-click to execute the first step

Design: Sci-fi/multiverse aesthetic — dark background, glowing timeline paths branching from central decision point.

### Step 5: Execute the Chosen Path
- **Mode:** deterministic
- **Output:** Decision card

Options: "Choose Universe A — [first action]" / "Choose Universe B — [first action]" / "Choose Universe C — [first action]" / "I need more information before deciding"

## Output

An interactive multiverse visualization showing 3 alternate timelines branching from the decision point, each with probability-weighted outcomes, comparison matrix, historical precedents, and one-click execution of the first step.

## Lessons Learned

### Do
- Make each universe genuinely distinct — not just speed variations
- Ground probabilities in actual historical data when available
- Include the FIRST CONCRETE ACTION for each universe — avoid analysis paralysis
- Show reasoning behind probability estimates

### Don't
- Don't make one universe obviously the "right answer" — if obvious, CSM wouldn't need this
- Don't ignore the relationship dimension — sometimes "suboptimal" business choice is right for the relationship
- Don't present more than 3 universes — decision fatigue kills action

### Common Failures
- No historical precedents: fall back to first-principles reasoning and general best practices
- Decision too vague: ask for specific constraints and stakes
- All paths converge: tell the CSM the decision may not be as consequential as it feels

### When to Ask the User
- When stakes are unclear — "What's the worst thing that happens if you do nothing?"
- When missing key context — "Is there a renewal coming up? Exec relationship at play?"
- When all paths seem equivalent — ask what they value most (speed, relationship, revenue)