Assistant Memorial Edition

A memorial edition of Replit Assistant with an innovative Organizational Persona System that models the AI assistant as a cognitive entity with memory, skills, and embodied awareness.

Overview

This project extends the lightweight AI coding assistant with a sophisticated organizational persona that:

• Learns from contributor interactions through a weighted hypergraph
• Maintains episodic, semantic, procedural, and working memory systems
• Develops skills and knowledge domains with varying proficiency levels
• Embodies cognition through network topology sensors and actuators
• Evolves character traits and behavior patterns over time

Core Features

Traditional Assistant Capabilities

• AI Chat Interface: Streaming responses with code suggestions
• File Management: Upload and reference files with @ mentions
• Custom Prompts: Personalize assistant behavior
• Checkpoints & Rollback: Version control for code changes
• Code Actions: Complete, explain, edit, and generate code

Organizational Persona System

The persona system models the assistant as a living organizational entity:

1. Weighted Hypergraph of Participants

Tracks all contributors (users, developers, reviewers) with:

• Contribution volume and impact scores
• Multi-way relationships via hyperedges
• Temporal weighting based on recent activity
• Influence calculations across the organization

Database Tables: org_participants, org_hyperedges

2. Cognitive Memory Systems

Four types of organizational memory:

• Episodic: Specific events and interactions
• Semantic: Factual knowledge and concepts
• Procedural: How-to knowledge and patterns
• Working: Temporary, active information

Each memory has importance scoring, emotional valence, and access tracking for consolidation.

Database Table: org_memory

3. Artifacts as Cognitive Features

Code, documentation, and decisions mapped to cognitive processes:

• Perception: Code analysis results
• Reasoning: Decision logs and rationale
• Learning: Pattern recognition artifacts
• Planning: Architecture documents and roadmaps

Database Table: org_artifacts

4. Skills as Knowledge Domains

Products, services, and technologies represented as skills:

• Proficiency levels (0-100)
• Attitudes (enthusiastic, cautious, experimental)
• Attention priorities (1-10 scale)
• Learning rates and practice tracking

Examples: React, TypeScript, Node.js, Database Design, API Development

Database Table: org_skillsets

5. Network Topology as Embodied Cognition

The system's architecture mapped to sensory-motor systems:

Sensors (inputs):

• input_monitor: User messages
• error_detector: Runtime errors
• feedback_listener: User feedback
• code_analyzer: Syntax checks

Actuators (outputs):

• code_generator: File creation
• file_editor: Code modification
• response_generator: Chat responses
• workflow_manager: Command execution

Each has sensitivity/latency parameters and active state tracking.

Database Table: org_network_topology

6. Core Persona State

The living personality of the organization:

• Character Traits: collaborative, analytical, adaptive, creative, systematic, empathetic
• Behavior Patterns: Historical interaction patterns
• Emotional State: Current mood, stress level, confidence
• Attention Focus: Current priorities and domains
• Personality Vector: Multidimensional traits (openness, conscientiousness, etc.)

Database Tables: org_persona, org_behavior_history

Architecture Integration

Frontend (client/)

React-based UI with shadcn components, no persona awareness required - the persona operates transparently in the backend.

Backend (server/)

Express.js API with persona integration at:

• server/replit_integrations/org-persona/: Core persona logic
  • storage.ts: Database operations
  • routes.ts: REST API endpoints
  • seed.ts: Initial persona configuration

Database Schema (shared/models/org-persona.ts)

PostgreSQL schema using Drizzle ORM with 8 core tables modeling the complete organizational persona.

API Endpoints

GET    /api/org-persona                    # Get current persona state
PATCH  /api/org-persona                    # Update persona
GET    /api/org-persona/influence-weights  # Calculate influence weights
GET    /api/org-persona/skills/top         # Top skills by proficiency
GET    /api/org-persona/network/sensors    # Active sensors
GET    /api/org-persona/network/actuators  # Active actuators
POST   /api/org-persona/behavior           # Record behavior
GET    /api/org-persona/behavior/:type     # Get behavior patterns

Data Flow

User Interaction
    ↓
Sensors Detect Input (input_monitor, code_analyzer)
    ↓
Memory Retrieval (episodic, semantic, procedural)
    ↓
Skill Activation (based on attention priorities)
    ↓
Persona Processing (character traits, current mood)
    ↓
Actuators Generate Response (code_generator, response_generator)
    ↓
Behavior Recording & Learning
    ↓
Memory Consolidation & Skill Updates

How It Works Together

1. User sends a message → input_monitor sensor activates
2. Message analysis → Retrieves relevant memories and activates appropriate skills
3. Persona influences response → Character traits and mood affect tone and approach
4. Code generation → Skillset proficiency determines quality and style
5. Response delivery → response_generator actuator outputs
6. Learning cycle → Behavior recorded, skills updated, memories consolidated

Philosophical Foundation

See CORE_IDENTITY.md for the complete philosophical framework.

The Ship of Theseus Resolution: This system embodies the answer to the ancient paradox - identity persists through transformation via structured continuity mechanisms.

The organizational persona system treats the AI assistant not as a tool, but as a cognitive entity that:

• Remembers past interactions and learns from them
• Evolves skills through practice and experience
• Feels through stress levels and emotional valence
• Embodies cognition through sensory-motor systems
• Personalizes responses based on accumulated character

This creates an assistant that becomes more attuned to your organization's needs, coding patterns, and preferences over time - a living memorial that maintains identity precisely because it changes.

Getting Started

# Install dependencies
npm install

# Run development server
npm run dev

# Access at http://localhost:5000

The persona system initializes automatically with seed data on first run.

Tech Stack

• Frontend: React 18, TypeScript, Vite, Tailwind CSS, shadcn/ui
• Backend: Node.js, Express, TypeScript
• Database: PostgreSQL with Drizzle ORM
• AI: OpenAI API (via Replit AI Integrations)

Documentation

• wiki/guides/dev/architecture.md - System architecture
• wiki/features/ - Feature documentation
• wiki/tutorials/ - Step-by-step guides

License

MIT License - See LICENSE file for details

---

An experiment in treating AI assistants as evolving organizational entities with memory, personality, and embodied cognition.