Implementation Overview: AI vs Human vs Hybrid

“Know when to use the tool. Know when to be the tool.”

Every process in Liana Banyan has a defined implementation type. This transparency ensures everyone knows where AI assists, where humans lead, and where both collaborate.

The Three Implementation Types

┌─────────────────────────────────────────────────────────────┐
│                    IMPLEMENTATION TYPES                      │
├─────────────────────────────────────────────────────────────┤
│                                                              │
│  🤖 AI IMPLEMENTATION                                        │
│  ══════════════════                                          │
│  Automated, no human intervention                            │
│  Examples: Vote counting, math calculations                  │
│                                                              │
│  👤 HUMAN IMPLEMENTATION                                     │
│  ═══════════════════════                                     │
│  Manual, requires human action                               │
│  Examples: Final approvals, moderation decisions             │
│                                                              │
│  🤝 HYBRID IMPLEMENTATION                                    │
│  ════════════════════════                                    │
│  AI proposes, human reviews (or vice versa)                  │
│  Examples: Patent drafting, letter writing                   │
│                                                              │
└─────────────────────────────────────────────────────────────┘

Complete Implementation Map

🤖 AI-Only Processes

Characteristics:

• Zero human intervention
• Deterministic outcomes
• Auditable logs
• Real-time execution

👤 Human-Only Processes

Characteristics:

• Requires judgment
• Context-sensitive
• Accountability assigned
• Cannot be automated

🤝 Hybrid Processes

Characteristics:

• AI handles volume
• Human handles judgment
• Clear handoff points
• Documented workflow

The Hybrid Workflow Patterns

Pattern 1: AI Proposes, Human Disposes

AI generates output
       ↓
Human reviews
       ↓
  ┌────┴────┐
  │         │
APPROVE   REJECT
  ↓         ↓
Proceed   Return to AI
          (with feedback)

Examples:

• Patent claims
• Letter drafts
• Code changes

Pattern 2: Human Creates, AI Verifies

Human creates content
        ↓
AI verifies (Star Chamber)
        ↓
   ┌────┴────┐
   │         │
 VALID    INVALID
   ↓         ↓
Proceed   Flag for review

Examples:

• Financial claims
• Technical specifications
• Public statements

Pattern 3: Parallel Processing

┌─────────────────┬─────────────────┐
│    AI TRACK     │   HUMAN TRACK   │
├─────────────────┼─────────────────┤
│ Generate draft  │ Review context  │
│       ↓         │       ↓         │
│ Self-verify     │ Prepare feedback│
│       ↓         │       ↓         │
└────────┬────────┴────────┬────────┘
         │                 │
         └────────┬────────┘
                  ↓
            MERGE POINT
           (Human decides)

Examples:

• Complex research
• Strategy development
• System design

Implementation by System

Star Chamber (AI Systems Trunk)

Voting System (Governance Trunk)

Three-Gear Currency (Economic Trunk)

Context Management (AI Trunk)

Decision Matrix: When to Use What

Use AI When:

• ✅ Task is rule-based
• ✅ Outcome is deterministic
• ✅ Volume is high
• ✅ Speed matters
• ✅ Consistency required
• ✅ No judgment needed

Use Human When:

• ✅ Legal liability involved
• ✅ Relationship matters
• ✅ Context is complex
• ✅ Values conflict
• ✅ Precedent unclear
• ✅ Accountability required

Use Hybrid When:

• ✅ AI can handle 80% of work
• ✅ Human needed for edge cases
• ✅ Quality assurance required
• ✅ Both skills contribute
• ✅ Workflow has natural handoffs
• ✅ Either could fail alone

Why Document Implementation Types?

Transparency

• Everyone knows what’s automated
• No hidden AI decisions
• Audit trail exists

Accountability

• Clear ownership per process
• Know who to ask
• Responsibility assigned

Improvement

• Identify bottlenecks
• Know what to automate next
• Measure AI vs human performance

Trust

• Members understand the system
• No “black box” feeling
• Predictable behavior

The MimicTrunk Integration

As members progress through MimicTrunk trust levels, their implementation access changes:

Higher trust = more AI autonomy.

Implementation Changelog

Track when implementations change:

Know the tool. Be the tool. Know when.

FOR THE KEEP!