Designing a Knowledge Hub for AI-Powered Workflows

The Problem: A Disconnected Knowledge Ecosystem

AI-powered workflows generate a lot of artifacts: chat sessions, prompts, code snippets, and conceptual notes (we call them "seeds"). But these artifacts are often disconnected, living in separate silos. This makes it hard to see the big picture and understand how ideas connect and evolve over time. We wanted to solve this problem by creating a unified Knowledge Hub for our 11-11 platform.

The Build: A Three-Tiered Experience

We designed a three-tiered experience to help users explore their knowledge ecosystem:

1. The Trail of Thought: A contextual timeline that shows the lineage of a specific artifact. Where did this idea come from? What other artifacts were created from it? This is the "micro view."
2. The Hub v2 (Activity Feed): A chronological, filterable feed of all knowledge creation and connection activity. This is the "macro view."
3. The Graph Visualization: A full-screen, interactive graph that allows users to explore their entire knowledge ecosystem visually. This is the "god-mode view."

We built this system using Next.js, TypeScript, and react-force-graph for the visualization. A new knowledge_links table in our PGlite database stores the connections between artifacts, and a set of API endpoints provides the data for the different views.

Key Insights

• Progressive Disclosure is Key: Don't overwhelm users with a complex graph visualization upfront. Start with a simple, contextual view (the Trail of Thought) and allow users to progressively disclose more complexity as they need it.
• Multiple Views for Multiple Mental Models: Different users think in different ways. Some prefer a chronological feed, while others prefer a visual graph. By providing multiple views, we can cater to different mental models and use cases.
• The Backend is the Foundation: The most important part of this system is the knowledge_links table. A solid data model is the foundation for any good knowledge management system.

The Result: A Connected Thinking Environment

The final product is a complete knowledge management system that allows users to see how their ideas connect and evolve over time. It transforms 11-11 from a collection of disconnected tools into a true thinking environment. The next step is to add more powerful filtering and search capabilities to the Hub.

The Takeaway: Connect the Dots

Don't just build tools for creating knowledge. Build tools for connecting it. The real value comes from the relationships between ideas, not just the ideas themselves.