inevitability-engine（必然性引擎）

The Inevitability Engine

A research protocol for discovering novel software businesses that become inevitable due to AI capability improvements.

Core Philosophy

Thesis: We're witnessing the first-ever inversion of the tool adaptation curve. Historically, humans adapted to tools faster than tools evolved. Now, tools (LLMs) evolve faster than humans can adapt. This creates a capability overhang that unlocks previously impossible business models.

Three forcing functions:

1. Context window explosion (4K → 128K → 2M tokens in 24 months)
2. Inference cost collapse (~90% reduction/year)
3. Tool-use reliability (function calling: 60% → 95%+ in 18 months)

Result: The "synthetic worker" isn't metaphor—it's infrastructure. Companies will hire, fire, eval, and SLA these entities. The opportunity lies in tooling, governance, coordination, and domain specialization of this new workforce layer.

Quick Start

What do you want to do?

1. Full discovery process → Continue to Core Workflow (execute all 6 phases)
2. Map AI capabilities → Jump to Phase 1: Capability Frontier Mapping
3. Find pain points → Jump to Phase 2: Opportunity Discovery
4. Generate business ideas → Jump to Phase 3: Business Model Generation
5. Validate opportunities → Jump to Phase 4: Market Validation
6. Score inevitability → Jump to Phase 5: Inevitability Scoring
7. Create deliverable → Jump to Phase 6: Synthesis & Output

Core Workflow

Phase 1: Capability Frontier Mapping

Goal: Understand what's possible now and what becomes possible at each time horizon (3mo, 6mo, 12mo, 18mo, 24mo).

Load references/capability-mapping.md for detailed protocol.

Quick execution:

1. Map current AI capabilities on Wardley evolution axis (Genesis → Custom → Product → Commodity)
2. Identify constraint removals (what was impossible 12 months ago that's trivial now?)
3. Project forward using scaling laws and roadmaps
4. Build capability unlock timeline

Key research queries:

• "GPT-4 capabilities vs GPT-5 predictions site:openai.com OR site:anthropic.com"
• "context window roadmap LLM 2024 2025"
• "agent orchestration frameworks production deployment"
• "inference cost trends 2024 2025"

Output: Capability timeline showing what becomes automatable at each horizon

Phase 2: Opportunity Discovery (Segment-Problem Analysis)

Goal: Build exhaustive matrix of segments × problems to find high-value automation targets.

Load references/opportunity-discovery.md for detailed protocol.

Target segments:

• SMBs (1-50 employees)
• Mid-market (51-500 employees)
• Enterprise (500-5000 employees)
• Megacorps (5000+ employees)
• Knowledge workers (writers, designers, programmers, engineers, managers, finance, legal, healthcare, educators, researchers)

For EACH segment, discover:

1. Top 10 time-consuming tasks
2. Top 10 frustrations with current tools
3. Information work bottlenecks
4. Manual workarounds
5. Budget allocated to solutions

Key research pattern:

• "[segment] biggest time wasters 2024"
• "[segment] workflow automation pain points"
• "[segment] AI adoption barriers"
• "site:reddit.com [segment] productivity challenges"

Output: Segment-problem matrix with 50-100+ pain points identified

Phase 3: Business Model Generation

Goal: Transform high-potential opportunities into concrete business models with synthetic worker roles.

Load references/business-model-generation.md for detailed protocol.

Process:

1. Define synthetic worker primitives (10 atomic job functions)

- Continuous Monitor - Research Synthesizer - Document Processor - Communication Coordinator - Compliance Auditor - Creative Collaborator - Knowledge Curator - Workflow Orchestrator - Analysis Generator - Relationship Maintainer

1. Cross with segments to generate business ideas

- Example: Research Synthesizer × Legal = AI-powered legal research assistant

1. Map to time horizons based on capability unlocks

- 3mo: Document workspace agents - 6mo: Research automation platforms - 12mo: Synthetic operations teams - 18mo: Executive co-pilots - 24mo: Synthetic departments

For each opportunity, define:

• Synthetic worker role & SLA
• Economic leverage (cost reduction multiplier)
• Eval framework
• Human-in-loop points

Output: 25-50 business concepts with role definitions

Phase 4: Market Validation

Goal: Validate demand, size markets, analyze competition, identify differentiation.

Load references/validation-refinement.md for detailed protocol.

For top opportunities:

1. Search existing solutions

- "[business idea] startup 2024" - "[business idea] AI tool" - Assess: AI-native or bolt-on?

1. Find buyer intent

- "[segment] looking for [solution]" (Twitter, Reddit, HN) - Count mentions, upvotes, engagement

1. Estimate TAM/SAM

- "[segment] market size 2024" - "[job function] salary [geography]" - Calculate: # workers × % replaceable × willingness to pay

1. Analyze competition

- What's their wedge? (product-led, sales-led, platform) - What's their constraint? (tech debt, sales cycle, capital) - What's the orthogonal attack?

Output: Validated opportunities with market sizing and competitive analysis

Phase 5: Inevitability Scoring

Goal: Quantify which opportunities are inevitable and when.

Load references/inevitability-framework.md for detailed formulas and examples.

Inevitability formula:

Inevitability = (Economic_Pressure × Technical_Feasibility × Market_Readiness) / Adoption_Friction

Where:
E = (current_cost / ai_cost) - 1  [scale 0-10]
T = % of workflow automatable  [scale 0-10]
M = (existing_budget + behavior_change_readiness) / 2  [scale 0-10]
F = integration_cost + trust_gap + regulatory_barrier  [scale 1-10]

Threshold: Score > 25 = inevitable within stated horizon

For each opportunity:

1. Calculate economic pressure (cost ratio)
2. Assess technical feasibility (% automatable)
3. Gauge market readiness (budget + willingness)
4. Estimate adoption friction (barriers)
5. Compute score
6. Rank by inevitability

Output: Ranked list of opportunities with inevitability scores

Phase 6: Synthesis & Output

Goal: Create structured deliverable with actionable insights.

Load references/output-templates.md for formatting examples.

Standard deliverable structure:

1. Executive Summary (2 pages)

- Capability trajectory overview - Top 10 opportunities by inevitability score - Recommended actions

1. Opportunity Matrix (spreadsheet/table)

- 25-50 businesses ranked by horizon and score - Segment, problem, solution, economics, competition - Time to revenue estimates

1. Deep Dives (5-10 pages each, top 5 opportunities)

- Market analysis - Technical feasibility - Business model canvas - Go-to-market strategy - Risk factors - SLA definitions

1. Research Appendix

- All search queries executed - Key sources and citations - Assumption log - Uncertainty flags

Output: Comprehensive research report ready for decision-making

Key Frameworks

Wardley Evolution Axis

Map capabilities across evolution stages:

GENESIS → CUSTOM → PRODUCT → COMMODITY
├─ Multimodal reasoning (custom→product)
├─ Long-horizon planning (genesis→custom)
├─ Reliable tool orchestration (product→commodity)
├─ Real-time learning loops (genesis)
├─ Inter-agent coordination (genesis→custom)
├─ Domain-specific fine-tuning (custom→product)
└─ Eval frameworks (custom→product)

Load references/wardley-mapping.md for detailed methodology.

Time-Horizon Capability Unlocks

Synthetic Worker Primitives

10 atomic job functions that become commoditized:

1. Continuous Monitor - Watches systems, alerts on anomaly
2. Research Synthesizer - Gathers sources, summarizes, cites
3. Document Processor - Extracts, validates, transforms
4. Communication Coordinator - Drafts, routes, tracks
5. Compliance Auditor - Checks rules, flags violations
6. Creative Collaborator - Generates variants, iterates on feedback
7. Knowledge Curator - Organizes, tags, retrieves
8. Workflow Orchestrator - Manages multi-step processes
9. Analysis Generator - Runs reports, identifies patterns
10. Relationship Maintainer - Tracks context, personalizes outreach

Cross these with target segments to generate business ideas.

Research Protocol Patterns

Load references/research-protocols.md for complete query library.

Capability tracking:

• "GPT-5 capabilities predictions 2025"
• "Claude context window roadmap"
• "LLM tool use reliability production"

Pain point mining:

• "[segment] workflow inefficiencies reddit"
• "[segment] biggest productivity challenges"
• "[job function] time tracking studies"

Market validation:

• "[business idea] startup funding 2024"
• "[segment] software spending trends"
• "[task] automation ROI case studies"

Competitive intelligence:

• "AI [task] automation companies"
• "[competitor] customer reviews G2 Capterra"

First Principles Decomposition

For each high-value task:

1. Irreducible cognitive work?

- Reading, synthesizing, deciding, creating, coordinating?

1. % automatable TODAY?

- Use current LLM benchmarks (MMLU, HumanEval, etc.)

1. % automatable at each horizon?

- 3mo, 6mo, 12mo, 18mo, 24mo

1. What remains human-in-loop?

- Judgment, taste, stakeholder management, ethical choice

1. Economic leverage?

- Calculate: (human_cost - ai_cost) / ai_cost

Quality Signals

Good opportunity has:

• [ ] Economic pressure > 10x cost reduction
• [ ] Technical feasibility > 70% automatable within horizon
• [ ] Market readiness (existing budget + proven pain)
• [ ] Low adoption friction (easy integration, low trust gap)
• [ ] Clear SLA definition
• [ ] Measurable eval framework
• [ ] Validated buyer intent (social proof)
• [ ] Differentiated positioning vs incumbents
• [ ] AI-native architecture (not bolt-on)
• [ ] Workflow replacement (not just enhancement)

Red flags:

• Only 10-20% cost reduction (not compelling)
• High human-in-loop requirements (doesn't scale)
• Unclear eval criteria (can't measure success)
• Heavy regulatory burden (slow adoption)
• Strong incumbents with AI-native approaches
• No clear buyer intent signals
• Requires behavior change AND new budget

Execution Checklist

When running full discovery process:

• [ ] Phase 1: Capability Frontier Mapping (2-3 hours)
• [ ] Phase 2: Segment-Problem Discovery (8-10 hours, 15 segments)
• [ ] Phase 3: Business Model Generation (6-8 hours, top 25 opportunities)
• [ ] Phase 4: Market Validation (10-12 hours, top 50 opportunities)
• [ ] Phase 5: Inevitability Scoring (2-3 hours)
• [ ] Phase 6: Synthesis & Output (8-10 hours)

Total estimated research time: 40-50 hours

Can execute in iterations:

• Sprint 1: Phases 1-2 (discover landscape)
• Sprint 2: Phases 3-4 (generate and validate)
• Sprint 3: Phases 5-6 (score and synthesize)

Meta-Instructions

Prioritize businesses where:

• AI is native infrastructure, not bolted on
• 10-100x cost reductions, not 10-20%
• Workflow replacement over enhancement
• Synthetic workers are competitive advantage, not just efficiency

Constraints:

• No crypto/web3 businesses
• No consumer social (focus B2B, prosumer)
• No hardware-dependent models
• Prefer high-margin software (>70% gross margin potential)
• Prefer businesses that scale with inference, not headcount

Success criteria:

• At least 10 opportunities with inevitability score > 30
• At least 3 opportunities actionable within 90 days
• At least 1 opportunity worth spinning out as venture-backed startup
• Clear time-to-revenue estimates for each

Integration Points

With web research capabilities:

• Use WebSearch extensively for pain point mining
• Use WebFetch for detailed competitive analysis
• Use Grep for local codebase capability assessment

With other skills:

• process-mapper: Validate automation feasibility for specific workflows
• research-to-essay: Transform findings into thought leadership content
• strategy-to-artifact: Convert opportunity analysis into pitch decks

With business context:

• Flag opportunities with BetterUp synergy (internal tool → external product)
• Highlight Catalyst packaging potential (repeatable, teachable, scalable)
• Identify unfair advantages from domain expertise

Common Use Cases

Trigger patterns:

• "Find AI business opportunities in [industry]"
• "What becomes possible with 2M context windows?"
• "Map the synthetic workforce opportunity space"
• "Identify inevitable AI-native businesses"
• "Where can we apply AI to replace entire job functions?"
• "What workflows become automatable in 6 months?"
• "Validate this AI business idea"
• "Calculate inevitability score for [opportunity]"

Example execution:

User: "Find AI business opportunities in legal services"

Response:

1. Load Phase 2: Opportunity Discovery
2. Focus on legal segment
3. Execute pain point research queries
4. Build problem matrix
5. Map to synthetic worker primitives
6. Generate 5-10 business concepts
7. Validate top 3 with market research
8. Calculate inevitability scores
9. Deliver ranked opportunities with GTM strategies

Anti-Patterns

Don't:

• Chase 10-20% efficiency gains (not venture-scale)
• Bolt AI onto existing workflows (prefer replacement)
• Ignore adoption friction (score honestly)
• Skip competitive analysis (surprises kill startups)
• Assume capabilities without validation (use benchmarks)
• Create businesses requiring massive behavior change
• Focus on technology demos vs business models
• Ignore unit economics (must have path to profitability)

Do:

• Look for 10-100x cost reductions
• Design AI-native workflows from scratch
• Score inevitability rigorously
• Deep dive competitive landscape
• Validate capabilities with current benchmarks
• Find natural adoption paths
• Build real businesses, not features
• Model unit economics from day one

Success Metrics

Research succeeds when:

• At least 10 high-scoring opportunities identified (>30)
• Market validation confirms buyer intent
• TAM/SAM estimates are defensible
• Competitive analysis reveals clear wedge
• Time-to-revenue is realistic
• Technical feasibility validated with benchmarks
• Economic models show path to profitability

Business succeeds when:

• Inevitability score proves accurate
• Market adopts faster than projected
• Unit economics improve with scale
• Synthetic workers deliver promised SLAs
• Customers achieve 10x+ ROI
• Competition validates space
• Clear path to market leadership

Ready to discover what's inevitable?

Choose your starting phase above, or ask: "Run full inevitability engine research on [domain/segment/opportunity]"