Questions about Lojin SuperBrain.
MIDAZ Dev (midaz.dev) is a Jakarta-based AI startup building Lojin SuperBrain, a company brain for AI agents. Member of NVIDIA Inception Program. Not affiliated with other brands named Midaz.
What is Lojin SuperBrain?
SuperBrain is a company brain for AI agents. It captures decisions and knowledge from AI chats, Slack, voice notes, docs and web clippings, distills them into a typed knowledge graph, and serves that context to agents such as Claude Code, Cursor and Gemini CLI over MCP, with citations back to the source.
How is it different from a vector database or document search?
A vector database stores chunks. SuperBrain distills raw conversations into durable decisions, ideas, knowledge and artifacts, links them with typed relations (supersedes, depends_on, contradicts), flags contradictions for human review, and compiles approved decisions into SKILL.md files that agents can run.
Which agents and tools does it work with?
Any MCP-compatible client, including Claude Code, Cursor and Gemini CLI. Capture channels include Slack, Telegram voice notes, a Chrome extension, Obsidian, and exports from ChatGPT, Claude and Gemini.
Where is my data stored, and is it used to train models?
SuperBrain runs self-hosted today, so your vault stays on your infrastructure. API keys are bring-your-own and encrypted per tenant. We do not train models on customer data. A managed multi-tenant cloud workspace is in development.
Which models does SuperBrain use?
Gemini 3.5 Flash-Lite for distillation and Gemini Embedding 2 for embeddings by default, with optional NVIDIA NIM inference and NeMo Retriever embeddings and reranking. Keys are bring-your-own.
How can my team get access?
We are onboarding a small group of design partners: AI-native teams of 10 to 250 people who use agents daily. Email [email protected] to request early access.
Ready to build?
If you want a performance-first build (Astro + Cloudflare) with clean engineering handoff, let’s talk.
Deploy AI Architecture