xiReactor
A platform of pluggable AI cores. Brilliant is the first public core — live, open source, and in the community's hands. More on the way.
A platform of pluggable AI cores. Brilliant is the first public core — live, open source, and in the community's hands. More on the way.
xiReactor started in summer 2025 as a centralized platform of pluggable AI cores — a way to bridge and enrich data between the tools teams already depend on. CRM to CRM. Ticketing to docs. ETL-style flows with an AI layer doing the messy parts that brittle field-mapping never could.
The frame is adjacent to Zapier, Make, and n8n, but reshaped around cores rather than nodes. Each core is a self-contained capability — a knowledge base, a CRM sync, an enrichment pipeline, a coding scaffold — that can be added, removed, enabled, or disabled per client. The platform is the substrate; the cores are what actually do the work.
That's the concept. Here's where it actually stands today.
The platform is built one core at a time. Other cores have been developed privately for specific client integrations and aren't public. Brilliant is the first to be released to the community — live, Apache 2.0, and in people's hands now. The knowledge-base problem — shared, governed context for AI agents — kept surfacing across conversations, so that's the piece that shipped first publicly.
Next on deck for public release:
For now, Brilliant is the public surface. Factory is next.
A shared, governed knowledge base for teams running AI agents — institutional memory that people and agents read from and write to through the same interface.
Open source · Apache 2.0
If you're here about Brilliant, the Brilliant page is the right place to start. To talk about the broader xiReactor concept — other cores, the platform idea, where it might still go — send a note.
Prefer LinkedIn? Connect with me on LinkedIn