Hey Peerlist! 👋

Modern engineering teams are drowning in tool fragmentation. Architectural decisions get buried in Slack threads, product specifications live in Notion, issues are tracked in Jira, and implementations happen on GitHub.

When context is scattered across four different platforms, engineers waste hours hunting for information, re-asking questions, or making decisions based on outdated threads.

We built WorkContext to solve this problem by creating an open-source, AI-powered productivity workspace that acts as a unified context layer across your entire stack—without adding another proprietary silo or forcing you into expensive per-seat AI markups.

🌟 Key Features & Capabilities

• Universal Cross-Tool Context Search: Perform semantic search across connected Slack threads, Notion pages, Jira tickets, and GitHub issues. Get grounded AI responses with explicit inline citations pointing directly back to the original source thread or commit.

• Privacy-First BYOK (Bring Your Own Key): Maintain total control over your data and API costs. Connect directly to Google AI Studio (Gemini), OpenAI, Anthropic, OpenRouter, or custom self-hosted endpoints.

• Real-Time CRDT Collaboration: Work together seamlessly in multi-user documents powered by Yjs, Hocuspocus, TipTap, and BlockNote with sub-millisecond document sync and live cursor presence.

• Smart Task Extraction: Turn notes into action items instantly. Highlight key text or action items inside your notes to extract and assign structured tasks across your workspace.

• 100% Open Source & Self-Hostable: Complete transparency with full source code availability. Run it locally or deploy it on your own cloud infrastructure using Docker and PostgreSQL.

🛠️ Technical Architecture & Stack

We designed WorkContext to be lightweight, performant, and developer-friendly:

• Frontend: Next.js 16 (App Router), TypeScript, Tailwind CSS, TipTap, BlockNote

• Backend: Node.js, Express 5, Prisma ORM

• Database & Vector Search: PostgreSQL with `pgvector` for unified relational data and fast cosine-similarity vector retrieval

• Real-time Engine: Yjs + Hocuspocus WebSocket server for CRDT collaboration

• AI Orchestration: Multi-provider API abstraction layer supporting streaming responses and structured output generation

🚀 Why We Chose `pgvector` over Dedicated Vector Databases

Instead of spinning up separate vector database clusters (like Pinecone or Weaviate), we chose PostgreSQL + `pgvector` for three core operational benefits:

1. Transactional Consistency: User permissions (RBAC), document metadata, and vector embeddings live in the exact same database. When a document or integration is removed, vectors are deleted atomically.

2. Zero Extra Infrastructure: Keeps deployment simple for self-hosters—one database handles relational storage, full-text search, and vector embeddings.

3. Hybrid Relational-Vector Queries: Allows us to filter vector similarity searches by workspace ID and user permissions in a single SQL query.

💻 Links & Community

• Live Demo: https://www.workcontext.me

• GitHub Repository: https://github.com/marowa-labs/workcontext

We are actively building in public and would love your feedback, bug reports, feature requests, and GitHub stars! Let us know what you think in the comments below.