02Real-time · Multi-tenant · AI-assisted
Task Management System
Multi-tenant task management platform supporting real-time collaboration, semantic search and AI-assisted workflows.
- Collaboration over Socket.IO
- Real-time
- Background job types on queues
- 3
- Secured MCP access for AI assistants
- OAuth 2.1
Built with
- Angular
- NestJS
- Socket.IO
- Redis
- PostgreSQL
- RAG
- MCP
Highlights
Key capabilities, at a glance.
- 01
Real-time updates that scale horizontally for many concurrent users
- 02
3 background job types (emails, exports, task events) offloaded to queues
- 03
Semantic search — find tasks by meaning, not exact keywords
- 04
AI assistants access task data through an MCP server secured with OAuth 2.1
How it works
The architecture, step by step.
Interact with the live view, then read the full walkthrough of every stage.
Request path & background jobs
01 / 05 · Angular
Responsive PWA client with lazy loading and efficient rendering of large data sets.
Request path & background jobs
- 01
Angular
Responsive PWA client with lazy loading and efficient rendering of large data sets.
- 02
NestJS API
Multi-tenant API with server-side pagination, filtering and sorting.
- 03
PostgreSQL
Primary relational store for tenants, projects and tasks.
- 04
Redis
Caching layer that accelerates data loading.
- 05
Bull Queue
3 background job types — emails, exports and task events — run asynchronously.
Parallel WebSocket channel
- 01
Socket.IO
WebSocket gateway for collaboration across teams and projects.
- 02
Real-time updates
Instant notifications that scale horizontally for many concurrent users.
RAG over tasks
- 01
User Query
A natural-language question, e.g. “what's blocking the release?”
- 02
Embeddings
The query is converted into a vector embedding.
- 03
Semantic Search
Vector search finds tasks by meaning, not just exact keywords.
- 04
Relevant Tasks
The most relevant tasks are retrieved as context.
- 05
AI Assistant
The assistant answers grounded in the retrieved tasks.
Secure assistant access
- 01
AI Assistant
An AI assistant needs to read or act on application data.
- 02
MCP Server
A dedicated Model Context Protocol server exposes a controlled set of tools.
- 03
OAuth 2.1
Every request is authorised with OAuth 2.1 — the assistant only gets what the user grants.
- 04
Secure Tools
Narrow, audited tools instead of raw database access.
- 05
Business Data
Task data is returned, scoped to the authorised tenant and user.
Contact
Have an interesting problem?
Let's build something useful.