Since the private beta launch, development has focused on the most unglamorous but important work: making the InterBrain actually usable at scale.

Early testers with larger vaults (100+ DreamNodes) were hitting serious performance issues - slow startup, occasional crashes, and frustrating lag. The original architecture simply wasn't optimized yet for the large number of DreamNodes that real-world usage demands. 

Over the past month, two major releases shipped focused entirely on scalability: 

GPU-Native Rendering - DreamTalks now render directly on the GPU instead of going through the DOM. This alone provides 3-4x more node capacity before any slowdown. 

Dynamic Node Loading - The UI is now decoupled from the vault. A configurable maximum number of nodes populate the constellation view persistently, while the rest load dynamically as needed - for example, when navigating the Liminal Web, relevant 

nodes fly in on demand. This makes the difference between a vault that crashes and one that loads smoothly. 

Direct Media Streaming - The old base64 encoding approach that was eating memory has been eliminated. Media now streams directly from disk. 

Smart State Management - The plugin now tracks vault state and skips unnecessary rescans. Refreshing (Cmd+R) no longer triggers a full reload - only what's changed gets updated. 

The result: the InterBrain plugin itself is no longer the performance bottleneck. For very large vaults, Obsidian's own vault scanning on startup still adds some seconds - that's outside plugin control - but the InterBrain layer now handles its work 

in the blink of an eye. 

A smaller but meaningful addition: search now combines fuzzy matching with semantic search. Fuzzy search is snappy and intuitive when you know what you're looking for (type a friend's name, find it instantly). Semantic search shines with longer 

queries - like finding relevant DreamNodes from a conversation transcript. The hybrid approach gives the best of both worlds, and at the end of the day, you don't need to think about it. It just works. 

Recently I had the chance to walk my friend Andrew through the process of getting started with the InterBrain on macOS.

We also walked through the first steps of using the essential features of the system and had some time to discuss the road ahead for the project 

Demo & Q&A:

Development is currently focused on Holarchy navigation UI. The existing Liminal Web shows connections between dreams and dreamers - horizontal navigation across your relational field.
The upcoming Holarchy mode will show sub/supermodule relationships between DreamNodes - vertical navigation up and down the hierarchy.
These two modes will complement each other: Liminal Web for relational exploration, Holarchy for structural depth.

This is also coming with a complete rewrite of the UI orchestration layer, echoing the same pattern as the recent optimizations - eliminating race conditions and making the whole system more reliable and robust. 

Your ongoing support makes this journey possible. Every contribution helps dedicate more time and energy toward bringing this vision into the world. 

If you'd like to support continued development, consider becoming a backer on Open Collective. Every contribution - whether one-time or recurring - directly funds development time.