Skip to main content

Project Planning

RavenmaskOS v4 development roadmap and feature tracking.

Overview​

RavenmaskOS has evolved from a personal productivity system to an AI Agent Platform with true domain intelligence and learned routing. The v4 plan represents a strategic pivot toward building a commercial product while dogfooding it for personal use.

Linear Project: RavenmaskOS - AI Agent Platform

Mission: Build a company around an AI agent platform with true domain intelligence and learned routing, dogfooding it first to prove value.

v4 Architecture​

Core Innovations​

InnovationDescription
Domain Intelligence Schema (DIS)Formal schema for modeling business/life domains that agents operate within
Raven ModelCognitive architecture with distinct data planes: Huginn (state), Muninn (memory), Context (identity), Domain (world)
Neural LLM RoutingTreating agent workflows as neural networks with learned routing and pathway reinforcement

Key Principles​

  1. UI-First Configuration - Everything configurable via Agent Studio UI, zero SSH/config files
  2. Domain Intelligence First - Agents must know their world before acting in it
  3. Learned Routing - Pathways that succeed get reinforced, failures decay
  4. Centralized User Profile - One place for identity, memory, permissions, preferences
  5. n8n as First-Class - 400+ integrations as sales weapon, executives configure dont customize
  6. Human in the Loop - Approval workflows for critical actions with case management

Implementation Phases (v4)​

PhaseNameDurationStatusLinear
0Validate & Stabilize1 week📋 TodoRAV-192
1Domain Intelligence Foundation2 weeks📋 BacklogRAV-197
2Memory & Context Layer2 weeks📋 BacklogRAV-204
3Visual Builder + CI/CD2 weeks📋 BacklogRAV-209
4Learned Routing2 weeks📋 BacklogRAV-210
5Deployment & Metrics2 weeks📋 BacklogRAV-211
6First CustomerMonth 3📋 BacklogRAV-212

Timeline: ~12 weeks to production-ready (Phases 0-5), Month 3 for first customer

Phase Details​

Phase 0: Validate & Stabilize​

Goal: Validate existing Norns agent works end-to-end before building new features.

Key Tasks:

  • Test task creation via Slack
  • Test task query functionality
  • Test calendar integration
  • Test Linear integration
  • Fix any blocking bugs

Exit Criteria:

  • All core Norns functions work reliably
  • Baseline established for Phase 1 work

Phase 1: Domain Intelligence Foundation​

Goal: Build the DIS foundation - the map that agents use to understand their domain.

Key Deliverables:

  • Neo4j container for structural memory
  • NATS JetStream for real-time events
  • DIS database schema
  • Domain Modeler UI (/domains/*)
  • Visual ERD canvas with React Flow
  • DIS Resolver Service API
  • RavenmaskOS domain dossier

New Infrastructure:

ServicePurpose
Neo4j 5Knowledge graph, structural memory
NATS 2.10Real-time event streaming

Phase 2: Memory & Context Layer​

Goal: Build the Raven Model memory architecture with User Profile Hub, RBAC, and HITL.

Key Deliverables:

  • Structural Memory (Neo4j sync with DIS)
  • Context Service (Huginn + Muninn + Domain assembly)
  • Event Fabric Integration (NATS)
  • Memory Consolidation Jobs
  • Memory Explorer UI (/memory/*)
  • User Profile Hub (/users/*)
  • RBAC System (roles, permissions)
  • Human in the Loop (/approvals/*, /cases/*)

HITL Features:

  • Approval workflow definitions
  • Tool approval requirements
  • Case management with timelines
  • Slack/email notifications
  • Full audit logging

Phase 3: Visual Builder + CI/CD​

Goal: Complete agent development experience with n8n as first-class platform.

Key Deliverables:

  • Visual Canvas (React Flow)
  • Prompt Editor with DIS variables
  • Tool Assignment with DIS triplet binding
  • Branch & Version Control
  • Agent Playground with execution trace
  • n8n Workflow Management (/workflows/*)
  • Integration Library UI (400+ integrations)

Phase 4: Learned Routing​

Goal: Implement Neural LLM routing - pathways that succeed get reinforced.

Key Deliverables:

  • Learning Service (weight initialization, updates, decay)
  • Routing Integration (weighted selection)
  • Feedback Loop (success/failure signals)
  • Pathway Dashboard
  • A/B Testing Framework

Phase 5: Deployment & Metrics​

Goal: Production-ready deployment pipeline with comprehensive metrics.

Key Deliverables:

  • CI/CD Deployment Pipeline
  • Merge Request Workflow with visual diff
  • Metrics & Analytics Dashboards
  • Complete Audit Trail
  • Executive Dashboard

Phase 6: First Customer​

Goal: Land first paying customer with multi-tenant deployment.

Key Deliverables:

  • Terraform AWS Infrastructure
  • Self-service Onboarding
  • Pricing Model & Billing
  • Sales Enablement Materials
  • Support System

Workflow Process​

Each feature follows a lifecycle tracked via Linear labels:

stage: Design → stage: Build → stage: Test → stage: Approval → Done

Labels Structure:

  • Phase labels: v4: Phase 0 - Validate through v4: Phase 6 - Customer
  • Stage labels: stage: Design, stage: Build, stage: Test, stage: Approval
  • Component labels: component: DIS, component: Neo4j, component: NATS, component: Bifrost, component: Studio, component: n8n, component: HITL, component: Norns

Tech Stack (v4)​

LayerTechnology
Graph DatabaseNeo4j 5
Event FabricNATS JetStream
Relational DBPostgreSQL 16
CacheRedis
API GatewayBifrost (custom)
UIAgent Studio (Next.js)
Workflowsn8n
AuthZitadel SSO
AgentNorns (LangGraph)
ObservabilityGrafana, Loki, Tempo

8 Life Domains​

The domain model remains consistent from earlier versions:

DomainCodenameFocus Area
FinanceHrafnhoardBudget, investments, goals
WorkRavenhelmCareer, professional
FamilyIdunns GardenRelationships, family
HealthEirs VitalityFitness, wellness
CreativeBragis QuillProjects, learning
HomeMidgardProperty, maintenance
HouseholdFriggs HearthDaily operations
DigitalMimirs LegacyDigital assets

Documentation​

See Also​