AI-Human Admin Dashboard

Tracking ODCI - On-Device Conversational Intelligence Assistant

Project Overview

A dual-interface project management dashboard for tracking development of ODCI — a hybrid edge-cloud AI system that captures meeting audio on-device (privacy-first), extracts entities and relationships, builds cumulative domain knowledge, and answers questions grounded in meeting history using GraphRAG.

Human Interface

Visual UI with charts, Kanban boards, architecture views, and metrics tracking

AI Interface

JSON state block in DOM for programmatic access by AI assistants

Dashboard Views

1 Project Overview Dashboard

Main dashboard with progress metrics, active tasks, microservices status, and resource usage

Dashboard View

2 System Architecture Blueprint

Interactive 5-layer AI stack visualization with GraphRAG data flow pattern

Architecture Blueprint

3 Phased Roadmap (Kanban)

Three-phase development roadmap: MVP Core Note-Taker, Domain Learner, Expert Assistant

Phased Roadmap

4 Governance & Evaluation

Performance metrics tracking: WER trends, DER rates, system latency, and privacy compliance

Evaluation & Metrics

5 Data & Model Management

Task management, backlog tracking, annotation pipeline, and knowledge graph ontology

Data & Models

Technology Stack

Single-File HTML

No build step required

Tailwind CSS

Modern utility-first styling

Vanilla JavaScript

No framework dependencies

ODCI 5-Layer AI Stack

Layer 5 Monitoring & Analytics Evaluation, Cost, Privacy
Layer 4 Orchestration & Agents App Server, RAG Coordinator
Layer 3 Intelligence (Models) STT, Diarization, NLP, LLM
Layer 2 Data & Memory Knowledge Graph, Vector DB
Layer 1 Infrastructure On-Device + Cloud

Key Features

Privacy-First

Raw audio stays on-device, only processed data goes to cloud

GraphRAG

Vector search + Knowledge Graph for factually grounded answers

localStorage

Simple persistence with JSON export for backup

AI-Readable

Embedded JSON state for programmatic AI access