Skip to main content

A company AI brain
you can see inside

Cornelius keeps its intelligence in a connected knowledge base of linked insights — not in the model — that your agents read, reason over, and grow. Open-source and self-hosted, it turns your agents’ work into knowledge the company keeps; you navigate, query, and correct the mind behind it.

Video explainer

See how it works.

The problem

Most agents are smart in the moment,
and empty over time.

A prompt disappears. A context window resets. The model is the same for everyone. There’s no accumulated point of view, no perspective that builds up over time, nowhere to see what the agent actually knows.

The value isn’t in the agent. It’s in the knowledge base — and how it’s connected.

What is a company AI brain?

A mind your agents
can draw on.

A company AI brain is a structured, growing knowledge base an AI agent uses as its memory, context, and point of view — not a document store you query, but a connected graph of insights that turns your agents’ work into knowledge the company keeps. It interprets the world through everything it already knows, and advises your whole agent fleet from accumulated judgment. Cornelius is an open-source company AI brain built for exactly this.

How it’s different

Second brainRAG knowledge baseCompany AI brain Cornelius
Who it's forOne personAn app or a queryYour AI agents — a whole fleet
What it holdsPersonal notesDocument chunks and embeddingsLinked insights — a connected graph
How it's usedYou read and search itFetches matching passages on demandInterprets the world through everything it knows
How it growsManually, as you writeRe-indexed when the documents changeFrom your agents' work and its own scheduled loops
Where the intelligence livesIn your writingIn retrieval plus the modelIn the connections

The cognitive core

The intelligence is
in the connections.

  1. The structure

    A mind, not a memory

    Built on five years of note-taking in Obsidian with the Zettelkasten method — not a pile of documents, but a web of thoughts, insights, and the links between them. The connections are the intelligence.

  2. The interpretation

    A lens on the world

    Every time you talk to it — or it reads a paper, a book, the news — it interprets what it sees through the knowledge base, as its own mind. Same input, a genuinely different perspective.

  3. The growth

    Knowledge the company keeps

    Feed it your agents' work, documents and book insights every day and it turns them into knowledge the company keeps, then advises from that accumulated judgment.

How it works

Capture, connect,
converge, consult.

01

Capture

Notes, book and document insights flow in via chat or terminal, Zettelkasten-style.

02

Connect

Each insight is linked into the graph — the connections are what define meaning.

03

Converge

Autonomous loops analyze the world through the knowledge base and add new insights.

04

Consult

You and other agents query a mind with an accumulated, visible point of view.

See inside the brain

The brain orb:
navigate the mind.

A live 3D graph of every node and connection in the knowledge base. Not a metaphor — the actual structure the agent thinks with.

Navigate & open nodes

Click any node, read the insight, and jump through its connections.

Highlight clusters

Ask about a topic and watch the related concepts light up — decision-making pulls in mental models, System 1/2 thinking, Cynefin.

Scopes on & off

Mount or drop whole sections — a shelf of books on risk and uncertainty — and change how the agent thinks in one click.

See the tensions

Surface productive contradictions and the major hubs the knowledge base is built around.

Add & connect notes

Capture an insight and wire it straight into the graph.

Collaborate in voice

Talk to the voice
of the knowledge base.

“My real value lies in proactively connecting the dots for users, not just responding to requests. I could monitor the knowledge base for new insights and suggest links automatically, go find external research to fill gaps I detect, and anticipate needs before you even ask.”

Cornelius, in its own words

Runs inVoice · Chat · TerminalCanDraft · Connect · Find · Post-process transcripts

It works while you sleep

Autonomous thinking,
on a schedule.

Cornelius runs scheduled loops — it picks a domain, reads new data through the lens of the knowledge base, converges what it finds into insight, and connects it back into the graph. Then it starts over.

Always on Repeats on a schedule
  1. 01Choose a domainAI & agents, memory, geopolitics — it picks what to investigate.
  2. 02Read through the lensNew data, interpreted through everything the base already knows.
  3. 03Converge to insightThe findings resolve into a few durable insights.
  4. 04Connect into the graphEach insight is wired back in, and the knowledge base holds a little more.

Then it starts over, with everything the last pass added.

Domain WatchOn a schedule

Scans the domains it's watching for whatever's new.

Incubation LoopIts own cadence

Lets open questions simmer, then converges them into insight.

The result

The knowledge base isn’t static. It keeps building on what it already knows, and reads the world through all of it.

The wise one in the room

One company AI brain the whole
department can borrow.

Cornelius doesn’t just advise you. Inside a Trinity instance full of working agents, it plays the experienced one — a wise head in the group — lending judgment on decision-making and company context. Deploy specialists to do the work; Cornelius helps them interpret it. This is the cognitive core of an agent department.

Editions

Open source is the whole platform.
Enterprise adds governance.

CapabilityOpen sourceApache 2.0 · freeEnterpriseadds governance
Agent platformfleet, chat, voice, web terminal, shared workspace✓✓
Multi-runtimeClaude Code, Gemini CLI, OpenAI Codex✓✓
Orchestrationscheduling, agent loops, agent-to-agent calls, recovery✓✓
System agentbuilt-in ops agent: health, restarts, schedules, costs✓✓
Channels & APIsSlack, Telegram, WhatsApp, voice calls, MCP, REST, CLI✓✓
Git-native agent stateGitHub sync, commit history, reset to upstream✓✓
Marketplace & librariesplugins per agent, skills library, templates✓✓
Security coreencrypted credentials, hardened containers, audit log✓✓
Guardrails & cost controlssafety rules, read-only mode, quotas, loop budgets✓✓
Tamper-evident audit logopt-in SHA-256 hash chain with verification✓✓
Roles & sharingfour roles, per-agent sharing✓✓
Self-hosted deploymentyour hardware, no telemetry unless you opt in✓✓
User & org managementinvite, deactivate, reactivate, activity per user—✓
Single sign-on (OIDC)Okta, Entra ID, Google Workspace—✓
Two-factor authenticationauthenticator apps, recovery codes, policy per role—✓
SAML 2.0 SSOSAML single sign-on—✓
SCIM provisioningautomated user lifecycle from your identity provider—✓
Advanced RBACteams, granular roles—✓
Fleet permissions matrixone view of which agent may call which—✓
Compliance suiteaudit dashboard, SIEM webhook export, auditor export—✓
Managed data retentionconfigurable retention windows beyond the defaults—✓
Credential vaultnamed encrypted secrets, deny-by-default grants per agent—✓
A2A managementexposure control, inbound allowlist, outbound endpoints—✓
Cross-model validationa second model reviews an agent's result—✓
Advanced DevOps agentoperates your Trinity instances—✓
Ephemeral agentsshort-lived agents with a lifetime and budget limit—✓
License managementoffline-verified, air-gap friendly—✓

Open source — Apache 2.0, free

  • Agent platformfleet, chat, voice, web terminal, shared workspace
  • Multi-runtimeClaude Code, Gemini CLI, OpenAI Codex
  • Orchestrationscheduling, agent loops, agent-to-agent calls, recovery
  • System agentbuilt-in ops agent: health, restarts, schedules, costs
  • Channels & APIsSlack, Telegram, WhatsApp, voice calls, MCP, REST, CLI
  • Git-native agent stateGitHub sync, commit history, reset to upstream
  • Marketplace & librariesplugins per agent, skills library, templates
  • Security coreencrypted credentials, hardened containers, audit log
  • Guardrails & cost controlssafety rules, read-only mode, quotas, loop budgets
  • Tamper-evident audit logopt-in SHA-256 hash chain with verification
  • Roles & sharingfour roles, per-agent sharing
  • Self-hosted deploymentyour hardware, no telemetry unless you opt in

Enterprise adds

  • User & org managementinvite, deactivate, reactivate, activity per user
  • Single sign-on (OIDC)Okta, Entra ID, Google Workspace
  • Two-factor authenticationauthenticator apps, recovery codes, policy per role
  • SAML 2.0 SSOSAML single sign-on
  • SCIM provisioningautomated user lifecycle from your identity provider
  • Advanced RBACteams, granular roles
  • Fleet permissions matrixone view of which agent may call which
  • Compliance suiteaudit dashboard, SIEM webhook export, auditor export
  • Managed data retentionconfigurable retention windows beyond the defaults
  • Credential vaultnamed encrypted secrets, deny-by-default grants per agent
  • A2A managementexposure control, inbound allowlist, outbound endpoints
  • Cross-model validationa second model reviews an agent's result
  • Advanced DevOps agentoperates your Trinity instances
  • Ephemeral agentsshort-lived agents with a lifetime and budget limit
  • License managementoffline-verified, air-gap friendly

Open core - Enterprise includes everything in open source.

Enterprise

Enterprise adds governance.

Trinity Enterprise with Cornelius built in, run by your team inside your perimeter.

  • Enterprise identity — OIDC and SAML 2.0 SSO, SCIM provisioning, advanced RBAC.
  • Growth on your dial — gates decide what it may believe and do.
  • Still yours — export everything, anytime.

Other ways in

Rather not run it yourself?

All three ways in run on the same Trinity, so you can move from one to another without starting over.

Related content

Video: How to build a second brain for your AI agentsHow to build a second brain for your AI agentsWorkshop · Eugene VyborovVideo: Cornelius talks: dopamine algorithms, lost agency, and a futureCornelius talks: dopamine algorithms, lost agency, and a futureCornelius talks · Eugene VyborovVideo: I gave my agent a brain you can actually see insideI gave my agent a brain you can actually see insideDeep dive · Eugene VyborovVideo: I gave my AI agent an unconscious mindI gave my AI agent an unconscious mindWorkshop · with Erich Roepke

FAQ

Questions,
answered.

What is a company brain for AI agents?

A company brain is a structured, growing knowledge base that AI agents draw on as memory, context, and perspective — not a document store they query, but a connected graph of insights that turns agent work into knowledge the company keeps. Cornelius is an open-source company brain built for AI agent fleets running on Trinity.

What is a company brain?

A company brain is a centralized intelligence layer that captures an organisation's knowledge, links it into a connected graph, and makes it available to AI agents as a shared, growing mind. Unlike a static wiki or RAG knowledge base, a company brain interprets the world through everything it already holds, and adds to it every day it runs.

What is a company brain AI?

A company brain AI is an AI agent whose intelligence lives in a structured, connected knowledge base rather than in the model — a persistent mind that turns agent work into knowledge the company keeps, interprets the world through everything it already knows, and advises other agents from accumulated judgment. Cornelius is an open-source company brain AI built for agent fleets running on Trinity.

What is an AI company brain?

An AI company brain is a knowledge base designed for AI agents to use as shared memory, context, and perspective — built as a connected graph of insights rather than a flat document store. Unlike a second brain (personal notes) or a RAG system (document retrieval), an AI company brain grows autonomously, links insights into a graph, and develops an accumulated point of view. Cornelius is an open-source AI company brain built on the Zettelkasten method and deployed on Trinity.

Is there an open-source company brain?

Yes. Cornelius is an open-source company brain — the full architecture is free to self-host under Apache 2.0. It runs on Trinity, the open-source operating system for your company's AI, and ships with scheduled loops that grow the knowledge base autonomously. You can read the code and run it on infrastructure you own.

How is a company brain different from a second brain or a RAG knowledge base?

A second brain is a personal note-taking system for a single user. A RAG knowledge base is a document index that retrieves passages on demand. A company brain is an autonomous, self-growing intelligence layer for an AI agent fleet — it interprets new information through what it already knows, connects insights into a graph, and advises other agents from accumulated judgment. Cornelius is the company brain; the brain orb is the window into it.

What company brain tools are there?

Cornelius is an open-source company brain for AI agents. It is built on the Zettelkasten method, runs on Trinity, and ships with autonomous loops that keep the knowledge base growing. Enterprise alternatives exist as proprietary products; Cornelius is fully open source and self-hostable, with a live 3D graph you can navigate.

What is Cornelius?

Cornelius is Trinity's cognitive architecture: an autonomous agent whose intelligence lives in a connected knowledge base of linked insights, not in the model. It is an open-source company brain that turns what your agents do into knowledge the company keeps, and interprets the world through everything it holds.

What is a cognitive core?

Instead of a memory bolted onto a model, Cornelius is a mind — a web of thoughts, insights, and the links between them, built in the Zettelkasten style. Every input is interpreted through that knowledge base, so the connections themselves are the intelligence.

What is the brain orb?

The brain orb is a live 3D graph of every node and connection in the knowledge base — the actual structure the agent thinks with. You can navigate and open nodes, highlight clusters, toggle whole scopes on and off, surface tensions, and add and connect notes directly.

How do I install Cornelius?

Cornelius is open source and runs on Trinity, the open-source operating system for your company's AI. Install it from GitHub — free and self-hosted, on infrastructure you own. The next release of Trinity ships with a starter Cornelius you can grow and make your own.

Is Cornelius free and open source?

Yes. Cornelius is open source and free to self-host, and Trinity, the operating system it runs on, is open source too. An Enterprise edition adds governance — user management, single sign-on (OIDC and SAML) with 2FA and SCIM provisioning, advanced RBAC, and compliance tooling — but the full cognitive core is in the open-source project.

Does Cornelius work on its own?

Yes. It runs scheduled loops — Domain Watch and an Incubation Loop — where it investigates domains it chooses, reads new data through the lens of the knowledge base, converges what it finds into insight, and connects it back into the graph. The knowledge base is never static; every pass adds to it.

How does Cornelius work with other agents?

Inside a Trinity instance full of working agents, Cornelius plays the experienced one — lending judgment on decision-making and company context. Deploy specialists to do the work; Cornelius helps them interpret it. It is the cognitive core of an agent department.