teamflow-tawny.vercel.app

TeamFlow

Production-grade AI-native team collaboration platform - a modern Slack/Microsoft Teams alternative built with cutting-edge technologies. Features real-time messaging with WebSocket infrastructure powered by Redis pub/sub and external Go service, multi-workspace support with organization-based isolation via Kinde Auth, and channel-based communication with threaded conversations. Includes AI-powered features using Grok 4.1 Fast: intelligent thread summarization with contextual understanding (English, Swahili, Sheng) tailored for Kenyan business context that identifies decisions, action items, and thread types; AI compose assistant for writing improvement, clarity enhancement, and structure optimization. Rich messaging capabilities with TipTap editor supporting markdown, code syntax highlighting via lowlight, image attachments through UploadThing, message editing/deletion, and emoji reactions with real-time updates. Built with Next.js 16 (React 19 with Compiler enabled), type-safe oRPC framework, PostgreSQL with Prisma ORM, Redis pub/sub messaging, Radix UI components with Tailwind CSS v4, TanStack Query for server state, and Arcjet security (WAF, multi-layer rate limiting, bot detection). Features comprehensive real-time presence with typing indicators, event-driven architecture with typed events (message CRUD, thread replies, reactions, presence updates), automatic reconnection with exponential backoff, optimistic UI updates throughout, infinite scroll message history, and end-to-end type safety with Zod validation. Performance-optimized with Turbopack, component caching, server-side prefetching, and enterprise-ready with professional error handling, SEO optimization, security hardening at multiple layers, and clean architectural patterns.

Intelligent collaboration

This platform is a production-grade Team Collaboration System built to rival Microsoft Teams and Slack, engineered with a specific focus on the Kenyan business context.

Core Architecture The system relies on a hybrid architecture using Next.js 16 (React 19) for the frontend and a dedicated Go service handling WebSocket connections via Redis Pub/Sub. This ensures sub-50ms message delivery and scalable presence management (typing indicators, online status).

AI-Native Implementation Unlike standard wrappers, this integrates Grok 4.1 Fast deeply into the workflow. The AI understands:

  1. Contextual Summarisation: Identifies decisions and action items in mixed-language threads (English/Swahili/Sheng).
  2. Smart Compose: An assistant that structures unstructured thoughts into professional business communication.

Security & Performance Security is handled via Arcjet (WAF, rate limiting) and Kinde Auth for multi-tenant organization isolation. The entire stack is fully type-safe from database to UI using Prisma and oRPC.

Customer:

Personal Project

Period:

2 months

Performance:

Full-Stack DevelopmentSystem ArchitectureAI/LLM IntegrationReal-time Infrastructure

Features:

AI IntegrationRedis Pub/Sub & Go MicroservicesEnd-to-End Type Safety (oRPC)Organisation-based Isolation

Background (Dark)

Brand Orange

Success Green

Text Muted

Scaling Real-Time Presence

Handling real-time typing indicators and status updates for thousands of users is resource-intensive. The Solution: We moved the WebSocket handling to a dedicated Go microservice. Using Redis Pub/Sub, we decoupled the state management from the Next.js serverless environment, ensuring the UI remains optimistic and snappy even under load.

TailFlow Main Chat Interface
AI Thread Summary Popover
Media unavailable

Would you like to be featured here as well?

Feel free to use the contact form or other contact options.

Request a Project