Architecture Misalignment Summary
Tier C — Historical. This document is a point-in-time snapshot preserved for reference. For current architecture, see CANONICAL_PLATFORM_ARCHITECTURE.md.
AGENTS.md vs. Current Implementation
The Gap
| Layer | Spec | Reality | Match |
|---|---|---|---|
| L0 | 4-stage onboarding | Login page only | ❌ 25% |
| L1 | 15 modules | 6 modules | ❌ 40% |
| L2 | 14 cognitive components | 5 components | ❌ 36% |
| L3 | 8-stage pipeline | 8 API modules | ✅ 100% |
Overall: 65% aligned with architecture
What's Missing
🔴 L0: Missing 75%
Spec: Entry → AI Insights → Demo → Context → Connect
Current: Just a login form
Need to Build:
/onboarding/welcome- AI personality analyzer/onboarding/demo- Live AI demo/onboarding/context- Path selector (Productivity Hub vs CS)/onboarding/connect- Tool connection wizard
🟠 L1: Missing 9 of 15 Modules
Spec: Home, Projects, Accounts, Contacts, Meetings, Docs, Tasks, Calendar, Notes, Knowledge, Team, Pipeline, Risks, Expansion, Analytics
Current: Dashboard, Accounts, Tasks, Calendar, Intelligence, Settings
Need to Build:
- Home (enhance current Dashboard)
- Projects - Cross-tool project tracking
- Contacts - Contact directory with relationships
- Meetings - AI-generated agendas
- Docs - Document hub with embeddings
- Notes - AI-assisted note-taking
- Knowledge - Knowledge graph
- Team - Team performance
- Pipeline - Deal tracking
- Analytics - Cross-module analytics
- Risks - Risk detection
- Expansion - Upsell opportunities
🟡 L2: Missing 9 of 14 Components
Spec: SpineUI, ContextUI, KnowledgeUI, Evidence, Signals, Think, Act, HITL, Govern, Adjust, Repeat, AuditUI, AgentConfig, DigitalTwin
Current: Signals (in Dashboard), Act (Actions), HITL (ProtectedRoute), partial Think (Intelligence), partial AuditUI (Settings)
Need to Build:
- SpineUI - Canonical data browser
- ContextUI - Unstructured content viewer
- KnowledgeUI - Knowledge graph explorer
- EvidencePanel - Evidence compiler ("Why this signal?")
- Govern - Policy enforcement UI
- Adjust - Feedback capture for learning
- Repeat - Continuous cycle visualization
- AgentConfig - Agent administration panel
- DigitalTwin - Simulation environment
The Figma Decision Tree
For UI/UX Designer:
Q1: L0 Onboarding - 4 screens or skip?
- ✅ Build all 4 - This is the "wow moment" per AGENTS.md
- Screen 1: AI personality insights
- Screen 2: Live demo (natural language → 4 tools)
- Screen 3: Context selection (Productivity vs CS)
- Screen 4: Tool connection
Q2: L1 Navigation - Flat or grouped?
- ✅ Grouped per AGENTS.md section 5
HOME
├── Dashboard
RELATIONSHIPS
├── Accounts
├── Contacts
WORK
├── Projects
├── Pipeline
├── Tasks
├── Calendar
├── Meetings
KNOWLEDGE
├── Docs
├── Notes
├── Knowledge Space
TEAM
├── Team
├── Analytics
INTELLIGENCE
├── Risks
├── Expansion
├── Insights
Q3: L2 Components - Where do they live?
- ✅ Overlay/panel pattern (not separate pages)
- L2SignalBar: Bottom bar (existing)
- EvidencePanel: Side drawer (new)
- SpineUI: Modal overlay (new)
- KnowledgeUI: Full-screen toggle (new)
- DigitalTwin: Full-screen simulation (new)
Q4: Domain Views - How to show 12 domains?
- ✅ Toggle buttons in Home dashboard
- Not separate pages
- Same page, different data view
- 12 toggle buttons: CS, Sales, RevOps, Marketing, Product, Finance, Service, Procurement, IT, Education, Personal, BizOps
Critical Decisions Needed NOW
Decision 1: Build L0 or Skip?
Options:
- A) Skip L0, use simple login (current) → Faster launch, less "wow"
- B) Build full L0 onboarding → 1-2 weeks, full AGENTS.md alignment
Recommendation: B for differentiation
Decision 2: All 15 L1 modules or MVP subset?
Options:
- A) Build all 15 now → 4-6 weeks, complete architecture
- B) Build core 6 (existing), add rest later → Launch faster, patch later
Recommendation: B for speed, with clear roadmap to 15
Decision 3: L2 as overlays or pages?
Options:
- A) Separate pages (/spine, /knowledge, etc.) → Simpler, less integrated
- B) Overlay panels (drawers, modals) → Complex, but AGENTS.md aligned
Recommendation: B for true L2 "layer" concept
Immediate Action Plan
Week 1: L0 Onboarding
- Design 4 onboarding screens
- Build AI insights mock
- Create tool connection UI
Week 2: L1 Core Enhancement
- Redesign sidebar navigation (grouped)
- Build Home dashboard (12 domain toggles)
- Create 3 new pages (Contacts, Projects, Pipeline)
Week 3: L1 Expansion
- Build 3 more pages (Docs, Notes, Knowledge)
- Build 3 more pages (Team, Analytics, Meetings)
Week 4: L2 Components
- Build EvidencePanel
- Build SpineUI overlay
- Build KnowledgeUI graph
Week 5: Polish
- Wire all L2 to L3 APIs
- Add DigitalTwin simulation
- Full integration testing
Bottom Line
Current: 65% aligned with AGENTS.md architecture
With L0 + 9 L1 + 9 L2: 100% aligned
Time to 100%: 4-5 weeks
Decision: Build L0 (critical for differentiation), add L1 modules incrementally, implement L2 as overlays.
The architecture is sound. The execution needs to match it.