Universal Memory for Every AI Agent · Live in Production
⚡ Accelerated by NVIDIA NIM & NeMo Retriever

Your company's knowledge,
executable by AI agents.

Universal memory for every AI agent. Lojin SuperBrain V3 captures scattered insights from AI coding sessions, voice notes, Obsidian, Slack, and web clips — distills them via a 6-agent council into a cited knowledge graph, and compiles them into live MCP skills your agents execute on demand.

100% Local-First Ready 15 Native MCP Tools AES-256 BYOK Multi-Provider Zero Silent Hallucinations
Vault Online
Omnichannel Inbound Stream SHA-256 Deduplicated
Claude Code & Antigravity MCP · Architecture Decision Distilled · 4 canonical nodes & edges linked
DEC-041
Telegram Voice Note · Founder GTM Strategy Distilling via Llama-3.3-70B / Gemini 2.5 Pro…
IDE-112
Obsidian 2-Way Vault Sync · Engineering SOPs Embedded (1536-dim) · Human Verified Trust
KNW-089
Chrome Web Clipper & Slack Thread · Competitor Audit Queued for Nightly Curator & Morning Briefing
ART-027
Edges: supersedes derived_from contradicts decided_by

Multi-Tenant Knowledge Graph

typed · cited · provenance-locked
Lojin SuperBrain Knowledge Graph Visualization
Scopes: Personal Vault Department Vault Company Brain
CEO Agent · 5D Strategic Evaluation NODE: IDE-2026-112
Autonomous MCP Skill Marketplace Evaluated against Company Constitutional Charter
9.3/10
Feasibility
9.2
Market Impact
9.6
Novelty
8.8
Unit Viability
9.1
Alignment
9.7
Decision: FAST-TRACK TO SKILL FACTORY Human Override Ready
Adversarial Council Debate 3 Personas + Arbiter
● Optimist (Growth Thesis)+High Leverage

Compiling recurring engineering fixes into MCP skills cuts onboarding time by 70% across all coding agents.

● Skeptic (Risk & Security)Guardrail Check

Unverified web clippings must never auto-compile into skills. Enforce minimum provenance trust = AI_SESSION.

★ Arbiter Synthesis VerdictApproved

Proceed with mandatory human-in-the-loop approval and provenance trust gating before publishing to MCP.

Stage 1 + Stage 2 Hybrid Retrieval Latency: 142ms
Query: "Why did we switch our static file serving and how is GPU reranking configured?"
Dense Vector (w=0.50) BM25 Lexical (w=0.30) Graph Walk 2-Hop (w=0.20) ⚡ NVIDIA NeMo Cross-Encoder (Logit +6.42)

Canonical containment verification (Path.resolve().is_relative_to()) was mandated to eliminate CWE-22 path traversal [DEC-2026-014]. Stage-2 GPU reranking uses nvidia/llama-3.2-nv-rerankqa-1b-v2 over RRF candidates with automatic local fallback [KNW-2026-089].

Hallucination Policy: Abstain when uncited Trust: HUMAN_VERIFIED
Retrieval Score Breakdown Reciprocal Rank Fusion
// HybridSearchService + Optional NVIDIA NeMo Reranker { "node_id": "DEC-2026-014", "provenance_trust": "human_verified", "rrf_score": 98.42, "score_breakdown": { "vector_rank": 1, "keyword_rank": 1, "graph_rank": 2, "nvidia_rerank_rank": 1, "nvidia_rerank_logit": 6.4218 } }
Skill Factory · Executable SKILL.md SKL-2026-019 (v2.4)
--- id: SKL-2026-019 name: zero-exposure-static-guard status: approved success_rate: 98.4% (62 runs) --- # Trigger: When building FastAPI static / file endpoints 1. Resolve canonical root: root = dist_dir.resolve() 2. Verify containment: target.is_relative_to(root) 3. Log execution outcome via superbrain_report_skill_run
Closed-Loop Telemetry: Auto-revises on failure
15 Native MCP Server Tools stdio & HTTP/SSE
// Connected to Claude Code, Antigravity & Cursor ▸ superbrain_capture // Ingest raw session ▸ superbrain_recall // Cited hybrid search ▸ superbrain_save_insight // Direct typed node ▸ superbrain_explore_relations// Graph traversal ▸ superbrain_verify_node // Promote trust tier ▸ superbrain_list_skills // Discover team SOPs ▸ superbrain_get_skill // Load executable skill ▸ superbrain_report_skill_run // Feedback telemetry
7 Live Capture Channels
6 Autonomous Vault Agents
15 Native MCP Tools
100% Source-Cited Answers
0 Silent Failures (Full Ledger)
Plugs directly into your stack
Claude Code MCP Google Antigravity Cursor & Windsurf Obsidian 2-Way Sync Telegram Voice Bot Slack Workspace Chrome Web Clipper NVIDIA NIM & NeMo GPU

Six specialized agents working your vault 24/7.

Most tools dump raw text into a vector database and hope for the best. SuperBrain runs a coordinated council of 6 specialized AI agents that structure, debate, verify, and compile your institutional memory.

AGENT 01 · DISTILLER Heavy Tier LLM

Canonical Distiller Engine

Transforms messy multi-turn transcripts, voice notes, and chat archives into typed nodes (Decision, Idea, Knowledge, Artifact, Process) while automatically linking contradictions and superseding outdated decisions.

Output: Typed Nodes + Edges ● Active
AGENT 02 · LIBRARIAN Zero-Hallucination RAG

Citation-Locked Librarian

Answers natural-language queries across Personal, Department, and Company vaults. Every claim carries mandatory inline citations to source nodes — and explicitly abstains when the vault lacks verified evidence.

100% Provenance Cited ● Active
AGENT 03 · CEO SCORER 5D Strategic Radar

CEO 5D Evaluator & Debate

Scores every captured idea across Feasibility, Market Impact, Novelty, Viability, and Alignment, then convenes a 3-persona adversarial debate (Optimist, Skeptic, Pragmatist) with an Arbiter verdict.

5D Radar + Arbiter Verdict ● Active
AGENT 04 · SKILL FACTORY MCP Compiler

Executable Skill Compiler

Compiles validated engineering and operational know-how into structured SKILL.md artifacts. Tracks real-world execution telemetry from coding agents and auto-revises skills when edge cases fail.

Serves via 15 MCP Tools ● Active
AGENT 05 · NIGHTLY CURATOR Autonomous Hygiene

Critic & Morning Briefing

Sweeps the graph every night to merge semantic duplicates, link orphan nodes, flag stale decisions, and generate an Executive Morning Briefing summarizing 24-hour organizational velocity.

Daily Executive Digest ● Active
AGENT 06 · SECRETARY Immutable Governance

Append-Only Audit Ledger

Logs every capture, distillation, BYOK verification, and MCP tool invocation in an append-only ledger with token counts, latency telemetry, and per-tenant USD cost attribution.

Zero Silent Failures ● Active

Local-first by default.
NVIDIA GPU-accelerated on demand.

Run 100% locally on SQLite + multi-provider BYOK (Gemini, OpenAI, Anthropic), or toggle optional NVIDIA NIM & NeMo Retriever microservices for TensorRT-LLM inference and Stage-2 Cross-Encoder reranking.

STAGE 01 · MULTI-SIGNAL FUSION

3-Signal Hybrid RRF Engine

Combines dense semantic vectors, exact lexical keyword matching, and 2-hop graph topology traversal into a unified Reciprocal Rank Fusion score.

  • Signal 1 Dense Cosine (w=0.50)
  • Signal 2 BM25 Lexical (w=0.30)
  • Signal 3 2-Hop Graph Walk (w=0.20)
STAGE 02 · OPTIONAL NVIDIA GPU

NVIDIA NIM & NeMo Retriever

Plug in an NVIDIA API key or self-hosted NIM container to unlock TensorRT-LLM completion and Stage-2 Cross-Encoder neural reranking with zero lock-in.

  • LLM NIM Llama-3.3-70B-Instruct
  • NeMo Embed nv-embedqa-1b-v2
  • NeMo Rerank nv-rerankqa-1b-v2
STAGE 03 · ZERO-TRUST GOVERNANCE

Provenance & Vault Isolation

Every node is scoped across 3 organizational boundaries and 4 cryptographic provenance trust tiers so unverified web clips never pollute core truth.

  • Encryption AES-256 Fernet / HKDF
  • Vault Scopes Personal · Dept · Company
  • Trust Tiers Human · AI · Team · Web
100% Local Fallback: Works offline without GPU keys Multi-Tenant: Strict row-level tenant isolation Markdown Mirror: Bi-directional Obsidian vault files
Request Architecture Whitepaper →

Beyond passive wikis and hallucinating RAG wrappers.

Static documentation rots the moment it is written, and generic RAG chatbots invent answers without accountability. SuperBrain closes the loop from capture to agent execution.

Capability Notion / Passive Wikis Standard Vector RAG Lojin SuperBrain V3
Knowledge Capture ✕ Manual copy-paste & formatting ✕ Static PDF / doc uploads only ✓ Live MCP, Voice, Slack, Obsidian & Chat Archives
Contradiction & Drift Handling ✕ Outdated pages silently coexist ✕ Retrieves conflicting chunks blindly ✓ Typed supersedes & contradicts graph edges
Search & Reranking Precision ✕ Basic keyword search ✕ Single-signal cosine similarity ✓ 3-Signal RRF + NVIDIA NeMo Cross-Encoder Rerank
Strategic Idea Evaluation ✕ None ✕ Generic chat opinion ✓ CEO 5D Radar + 3-Persona Adversarial Debate
Executable Agent Skills ✕ Humans must read & re-prompt ✕ Passive Q&A only ✓ Auto-compiled SKILL.md served live via 15 MCP Tools
Provenance & Cost Governance ✕ Page edit history only ✕ Uncited black-box generations ✓ 4 Trust Tiers + Append-Only Secretary Cost Ledger

From scattered chats to compounding intelligence.

Three movements. Capture becomes typed structure. Structure becomes strategic judgment. Judgment becomes executable skills your AI agents run automatically.

01

Connect your channels in 2 minutes

One-click register the MCP server into Claude Code, Antigravity, or Cursor. Link your Obsidian vault, Telegram voice bot, Slack workspace, or drag-and-drop full ChatGPT / Claude / Gemini export archives.

02

The 6-Agent Council structures & scores

The Distiller extracts decisions and SOPs; the CEO Agent scores ideas in 5D and runs multi-persona debates; the Nightly Curator deduplicates, auto-links graph edges, and prepares your Morning Briefing.

03

Agents recall & execute live skills

Query cited answers mid-session from any AI IDE, promote validated workflows into the Skill Factory, and let your agents load approved SKILL.md playbooks on demand. Every run feeds telemetry back into the brain.

Answers for founders & CTOs.

Everything you need to know about deployment, local-first privacy, MCP compatibility, and optional NVIDIA GPU acceleration.

Yes. SuperBrain is architected local-first out of the box with local SQLite + bi-directional Markdown mirrors, and supports Bring-Your-Own-Key (BYOK) encrypted with AES-256 Fernet or local self-hosted inference containers.
NVIDIA acceleration is completely modular. Without an NVIDIA key, SuperBrain uses your default local/BYOK setup. When you add an NVIDIA NIM API key (or point NVIDIA_BASE_URL to a self-hosted NIM container), SuperBrain unlocks TensorRT-LLM inference (Llama-3.3-70B) and Stage-2 NeMo Cross-Encoder reranking over hybrid search candidates.
Any Model Context Protocol (MCP) compatible host works natively—including Claude Code CLI, Claude Desktop, Google Antigravity, Cursor, and Windsurf. SuperBrain exposes 15 specialized tools for mid-session recall, insight capture, graph exploration, and skill execution.
Every node is tagged with a strict Provenance Trust Tier (Human Verified, AI Session, External Team, or Web Clipping). The Librarian Agent is governed by a strict citation prompt constitution: every statement must cite a specific node ID, and if no verified node answers the question, it explicitly abstains rather than guessing.

Ready to build a compounding brain for your team?

Leave your details and our engineering team will set up a tailored walkthrough with your own workflow, vaults, and AI agents.

✓ Full onboarding for Claude Code, Antigravity & Cursor MCP
✓ Personal, Department & Company multi-tenant vault setup
✓ Optional Local-First or NVIDIA NIM GPU-accelerated deployment

Or email us directly at [email protected]