1|# CyberAgent — Strategic Pivot: Building Long-term Moat
2|
3|> **“The universal personality layer for embodied AI.”**
4|
5|## Executive Summary: The Long Game
6|
7|**Current Reality (2026 Q2):**
8|- Behavior Tree.CPP dominates robotics (5-year head start, industry standard)
9|- Sony aibo has 30+ year brand + closed ecosystem (emotional lock-in)
10|- Commercial robot toys fragmented, no open platform
11|
12|**Strategic Pivot:** Shift from "open-source character gallery" to **"The Open Standard for Robot Personality."**
13|
14|### Core Insight
15|
16|**Behavior Trees are the assembly language of robot behavior — like x86 for robotics.**
17|
18|The winner won't be the best character gallery. The winner will be **the standard that defines how robot personalities are described, composed, and evolved over time.**
19|
20|### Moat Architecture (3-Layer Defense)
21|
22|```
23|┌─────────────────────────────────────────────────────────────┐
24|│ LAYER 1: Open Standard (The Foundation) │
25|│ • Character Description Format (CDF) — versioned JSON spec │
26|│ • Behavior Tree runtime (language agnostic) │
27|│ • Adapter interfaces (BLE, WebSocket, Serial, USB) │
28|└─────────────────────────────────────────────────────────────┘
29| ↓
30|┌─────────────────────────────────────────────────────────────┐
31|│ LAYER 2: Community Ecosystem (The Flywheel) │
32|│ • Marketplace: community characters, modules, adaptations │
33|│ • Tools: editor, simulator, telemetry studio │
34|│ • Contributors: 1000+ plugin creators │
35|└─────────────────────────────────────────────────────────────┘
36| ↓
37|┌─────────────────────────────────────────────────────────────┐
38|│ LAYER 3: Data Flywheel (The Killer App) │
39|│ • Real robot telemetry → improved character quality │
39|│ • Character evolution tracking across robot fleet │
40|│ • AI-powered character synthesis from behavior patterns │
41|└─────────────────────────────────────────────────────────────┘
42|```
43|
44|---
45|
46|## Competitive Landscape Analysis
47|
48|### Direct Competitors
49|
50|| Product | Type | Strength | Weakness | Our Wedge |
51||---------|------|----------|----------|-------------|
52|| BehaviorTree.CPP | Industrial BT framework | Mature, battle-tested | C++ only, steep learning curve, no character focus | **Developer experience + character design** |
53|| Sony aibo | Closed consumer product | Brand loyalty, emotional design | No open API, locked ecosystem | **Open standard, community-driven** |
54|| DJI RoboMaster | Educational robot kit | Hardware quality, SDK | Proprietary, limited personality depth | **Character-first SDK as addon** |
55|| Boston Dynamics Spot | Enterprise robot | Advanced locomotion | No personality layer, enterprise-only | **Consumer-friendly personality layer** |
56|
57|### **Critical Gap in Market**
58|
59|**No open platform bridges:**
60|1. **Character design** (creative, accessible) ↔ **Robot hardware** (physical, diverse)
61|2. **Community contributions** (crowdsourced personalities) ↔ **Production deployment** (robust, versioned)
62|3. **Behavior tree semantics** (modular, composable) ↔ **Emotional modeling** (lifelike, adaptive)
63|
64|This gap is where CyberAgent builds **irreversible network effects.**
65|
66|---
67|
68|## Long-Term Vision: 10-Year Path to Industry Standard
69|
70|### Phase 1: Foundation (Year 1-2) — Current
71|
72|**Goal:** Prove open-source platform has developer love.
73|
74|**Success Metrics:**
75|- [ ] 2,500 GitHub stars (up from ~120)
76|- [ ] 50+ community-contributed characters
77|- [ ] 10+ robot platform adapters
78|- [ ] First commercial integration (EDU or hobbyist kit)
79|
80|**Key Shift:** From "build features" to **"build community."**
81|```markdown
82|Q2 2026: Contributor program, first community characters
83|Q3 2026: SDK beta, character creator tool
84|Q4 2026: Marketplace beta, hardware partner program
85|Q1 2027: Official SDK v1.0, mobile companion app
86|```
87|
88|### Phase 2: Standardization (Year 3-4)
89|
90|**Goal:** Make CDF the de facto standard for robot personality.
91|
92|**Strategic Moves:**
93|1. **Publish CDF as open spec** (versioned, semantic versioning, backward compatibility)
94|2. **Open certification program:** "CyberAgent Compatible" badge for robot hardware
95|3. **Academic partnerships:** Publish research papers on embodied personality AI
96|
97|**Key Metric:** 3 major robot manufacturers adopt CDF in product SDKs.
98|
99|### Phase 3: Platform (Year 5-7)
100|
101|**Goal:** The community builds 10x faster than the core team.
102|
103|**Strategic Moves:**
104|1. **LLM-powered character synthesis:** Upload video of desired behavior → auto-generate BT
105|2. **Character evolution tracking:** See how "loyal-dog v2.3" differs from v1.0 across 10K robots
106|3. **Fleet intelligence:** Aggregated telemetry reveals character optimization patterns
107|
108|**Key Metric:** 80% of characters on platform are community-created.
109|
110|### Phase 4: Ecosystem (Year 8-10)
111|
112|**Goal:** Unquestioned industry standard.
113|
114|**Strategic Moves:**
115|1. **CDF becomes IEEE/W3C recommended standard**
116|2. **Nonprofit foundation governance** (like Linux Foundation model)
117|3. **Cross-domain adoption:** Entertainment, education, elderly care, service robotics
118|
119|**Ultimate Metric:** 10M+ robots running CyberAgent personalities worldwide.
120|
121|---
122|
123|## Immediate Action Items (Next Q2-Q3 2026)
124|
125|### 1. Community Program Launch (PRIORITY 1)
126|
127|```bash
128|Week 1-2: Contributor guide, first-timer issues, PR templates
129|Week 3-4: "First Character" tutorial series
130|Week 5-8: Open community character contest ($1K prize pool)
131|```
132|
133|### 2. SDK Architecture Decision (PRIORITY 2)
134|
135|**Decision:** Build **dual-language SDK** to maximize adoption:
136|- `@cyber-agent/core` (TypeScript/Node.js) — for web + edge devices
137|- `cyberagent` (Rust) — for embedded systems (ESP32, Raspberry Pi)
138|
139|**Why:** TypeScript leverages existing React ecosystem. Rust provides performance + memory safety for resource-constrained hardware.
140|
141|### 3. Character Creator Tool (PRIORITY 3)
142|
143|**Minimal viable product:**
144|- Drag-and-drop behavior tree builder
145|- Real-time simulation in browser (Three.js canvas)
146|- Export to JSON/CDF format
147|
148|**Differentiator:** No other competitor offers this level of visual BT editing for robot personalities.
149|
150|---
151|
152|## Risk Analysis & Mitigation
153|
154|| Risk | Probability | Impact | Mitigation |
155||------|-------------|--------|------------|
156|| BehaviorTree.CPP fork beats us | Medium | High | Faster developer experience, character-first positioning |
157|| Sony aibo opens SDK | Low | Medium | Establish standard before they react, focus on open ecosystem |
158|| Developer community doesn't engage | High | High | Gamification, monetary incentives, strong first PR |
159|| Hardware manufacturers resist standards | Medium | High | Start with hobbyist/EDU market, prove demand, then enterprise |
160|
161|---
162|
163|## Product Principles — Refocused for Long Moat
164|
165|1. **Standard First** — Every feature must advance the CDF spec
166|2. **Community-Owned** — Optimize for contribution, not control
167|3. **Hardware Agnostic** — Never lock to one robot platform
168|4. **Open Core** — Free forever, premium on services, never features
169|5. **Data Flywheel** — Telemetry must drive character improvements
170|6. **Developer Love** — DX > features in early years
171|
172|---
173|
174|## Quarterly Strategy Rhythm
175|
176|**Q2 2026:** Community launch + SDK architecture
177|**Q3 2026:** Character creator tool + marketplace beta
178|**Q4 2026:** Hardware partners +SDK v1.0
179|**Q1 2027:** Mobile app + AI character synthesis prototype
180|
181|**Each quarter asks:**
182|1. Did we move closer to becoming the standard?
183|2. Did we increase community contribution velocity?
184|3. Did we improve the data flywheel?
185|
186|---
187|
188|## Competitive Moat (Updated)
189|
190|- **Network Effects:** Each community character → attracts more robot builders → attracts more character creators
191|- **Standard Lock-in:** When you own the standard, switching costs become infinite
192|- **Data Moat:** 10 years of character behavior telemetry = unbeatable training data for character optimization
193|- **Contributor Flywheel:** 1000+ contributors = unkillable development capacity
194|
195|---
196|
197|*This document is the living strategy. Review quarterly with exec summary of progress.*
198|