🏗️ Architecture: What is the B2B2C multi-tenant tier model for the GoGrab AI ...

Date: 2026-02-05
Time: 23:02:45 UTC
Session: editor2-92.18
Agent: orchestrator
Confidence: 95%
Status: ✅ Documented

📋 Decision Context

Question:

What is the B2B2C multi-tenant tier model for the GoGrab AI platform?

Task: Session 93: Define B2B2C multi-tenant tier model

🎯 Decision Made

Chosen:

3-tier model: T1 (GoGrab devs/admins) → T2 (Space Builders — GoGrab's paying customers) → T3 (Platform Users — customers' customers). T2 is ONE tier: Space is a container (isolation boundary), Space Builder is the actor. They are entity vs actor, not separate tiers. Members[] on Space handles multi-builder scenarios with role-based access (owner/admin/editor/viewer).

Options Considered:

1. 4-tier model (separate Space and Builder tiers)
2. 3-tier model (Space=container, Builder=actor)
3. 2-tier model (GoGrab + everyone else)
4. Flat model with role-based access only

💭 Reasoning

The platform follows a B2B2C model. T1 are system operators (not in user_hierarchy — they use workspace-scoped data in ai-agents DB). T2 are L1 in user_hierarchy with subscriptionTier (free/builder/pro/enterprise). T3 are L2 in user_hierarchy, scoped to a universeId within a space. The key insight is that T2 is NOT two tiers — a Space is just a container, and a Space Builder manages it. Cross-space builder preferences live in ai-agents.builder_profiles (global), while per-space overrides live in space-agents_{spaceId}.preferences.

📁 Files Involved

• libs/libs-api/backend-core/src/lib/schemas/user-hierarchy.schema.ts
• libs/libs-api/backend-core/src/lib/schemas/platform-user.schema.ts

🔗 Related

• Decision ID: dev_1770332565808_97hq9ki
• Session: 92
• Workspace: editor2

Auto-generated by MCP Smart Consolidation