Try Studio
In Browserno installDesktop Mac AppDesktop Windows AppiPhone AppAndroid App
GitHub & Skills
Download Skill for ClaudeDownload Skill for OpenAI Codex
For
Myself — personal twinMy company — team memory

Your Company Brain

Your Company Brain shouldn't start empty.

Second Brain Link bootstraps institutional memory from the exports you already have — LinkedIn Company, Google Workspace, Slack, Notion, Confluence, Jira, Salesforce, HubSpot, Zendesk, email archives, Microsoft 365, Teams — into a structured, deduplicated, privacy-clean Company Brain your agents can use on day one. Emit to GBrain, or keep it in Obsidian for the team. Runs locally; employee PII is stripped in code.

building for yourself? Your personal brain →

Local-first·MIT·You're the data controller·Emits to GBrain
resolved employee

LinkedIn Company · Workspace · Slack · Notion · Jira · CRM · 12 sources → one Company Brain

Reads your exportsLinkedIn CompanyGoogle WorkspaceSlackNotionJiraSalesforceCRM exports+ more — 12 company sources
Emits toGBrainObsidianClaude Code & Codex
1
Prompt from exports to brain
0
Employee PII written to disk
100%
Local — you stay data controller
MIT
Your security team reads every line

Second Brain Studio

Browse and talk to your Company Brain — right in the browser.

No setup, no plugins. Open your Company Brain in a live workspace: explore how teams, people, and accounts connect, read notes, and ask an AI that reasons over the whole graph. Runs locally — your keys and data stay on your machines.

Interactive graph

A live, force-directed graph of every employee, team, customer and deal. Filter by type or source, search, and click any node to open its note.

Talk to your brain

A built-in assistant — Claude, OpenAI, Gemini and more — that cites the notes it used. Bring your own key; calls go straight to the provider. Nothing leaves your infrastructure.

Read, edit & track

A full Markdown editor with live preview, wikilinks and backlinks, plus a computed dashboard of your network and clusters.

Second Brain Mobile · iOS & Android

The same local-first brain, in every pocket.

A native mobile app that builds a personal twin entirely on-device — the same zero-network engine that powers the Company Brain. Your people's own exports become their own private brains on their own phones: no account, no backend, nothing touching your infrastructure.

Second Brain Mobile on iPhone — talk to your twin: warm intros, meeting prep, reconnectsSecond Brain Mobile on iPhone — talk to your twin (dark mode)
Second Brain Mobile on iPhone — the knowledge graph, rendered natively on the phoneSecond Brain Mobile on iPhone — the knowledge graph (dark mode)
Second Brain Mobile on iPhone — sources, imported and unpacked on-deviceSecond Brain Mobile on iPhone — sources (dark mode)

Built on-device

Exports are unpacked and transformed on the phone itself — the brain never exists anywhere else. Meeting prep and warm-path answers from the hallway, not the desk.

The graph in your hand

A native 3D graph of people, companies and topics — search it, filter it, tap into any note before you walk into the room.

Private by architecture

No account, no backend, zero telemetry. On-device AI on iPhone; on Android bring your own key — nothing for your security team to threat-model in the cloud.

On the App Store and Google Play.

The Agent Skill · Claude Code & OpenAI Codex

The engine itself — your agent builds the Company Brain.

The product that powers everything: the open-source engine, packaged as a cross-model Agent Skill. Install it once in Claude Code or OpenAI Codex and your agent knows how to turn export archives into a structured, deduplicated, privacy-clean Company Brain — on your infrastructure, with zero network calls.

your agent · with the second-brain-link skill
> "Build our Company Brain from exports/ — Slack, Workspace and the Salesforce dump" Skill: second-brain-link · 3 company sources detected · employee PII stripped in code Merging people across sources · building the graph… ✅ vault/acme-brain > "Who knows our billing system? Who owns the Acme deal?"

Runs where your data must stay

On a laptop, an air-gapped box, or your own server — dependency-free Python, zero network calls in the core. Nothing for security to threat-model in the cloud.

Self-adapting to your exports

Formats drift across tools and versions. The skill profiles every file, writes its own mapping rules, and rescues what it doesn't recognize — nothing is lost.

Emits your format

Obsidian vault by default, GBrain repo on request — plain Markdown either way, owned by you, readable by any agent.

The same skill powers Studio's reseed and the done-for-you onboarding — install it in one step or browse the source on GitHub.

The problem

The brain is the easy part now. Filling it isn't.

A new Company Brain is empty on day one, and your institutional knowledge is trapped in exports, shared drives, and old accounts. New hires re-learn what the company already knew. When someone leaves, the context leaves with them.

what your tools give you
exports/
├── workspace-export.zip       contacts, drive, 4.2 GB
├── slack-export.zip           channels.json, users.json
├── crm-accounts.csv           1,860 rows, stale owners
└── linkedin-company.zip       followers, employees

✕ four silos · same employee 4× · zero links
what your agents get
brain/
├── 00-org/company.md          org-rooted graph
├── 10-people/
   └── Dana-Reyes.md          1 person · 4 sources
├── 20-accounts/
   └── Northwind-Co.md        owner, history, stage
├── 40-projects/
└── _correlations/             people ↔ deals ↔ teams

✓ populated, deduplicated, privacy-clean — day one

Institutional memory that survives turnover — the context stays when people move on.

How it works

Point it at your exports. Get a connected Company Brain.

  1. 01

    Detect

    Point it at a folder of team exports — Workspace, Slack, Notion, Jira, Salesforce, HubSpot, LinkedIn Company… any of the 12 company sources. It catalogs every file, locally — and if your export's format drifts from the known shape, the skill adapts its own mappings to fit it, on your machines.

  2. 02

    Resolve

    Every employee, customer, and partner becomes one identity across sources — no more four copies of Dana.

  3. 03

    Build

    An org-rooted graph: the company at the center; people, accounts, deals, and projects linked around it. No AI calls.

  4. 04

    Emit

    Emit to GBrain for the team, or keep it as an Obsidian vault. Multiple entities get a cross-entity _correlations/ brain.

claude code · second-brain-link skill
> "Build our Company Brain from the exports in exports/" Detecting… workspace slack crm linkedin-co Resolving people across sources… 412 employees · 1,860 customers Building org-rooted graph… → _correlations/ + per-team brains ✅ brain/ · --emit gbrain to populate GBrain

Agent-ready

Institutional memory, as structured data.

20-accounts/Northwind-Co.md
--- type: account sources: [crm, workspace, slack] owner: "[[Dana Reyes]]" backup_owner: "[[Sam Patel]]" stage: expansion last_touch: 2026-05-28 projects: ["[[Rollout Q3]]"] --- # 18 months of history, from your own tools…

Org-rooted graph

The company sits at the center — 00-org — with employees, accounts, deals, and projects as typed, linked notes around it.

Ownership is explicit

Every account knows its owner and backup. "Who has the relationship?" is a graph query, not tribal knowledge.

Cross-entity correlations

Multiple teams or entities each get a brain, plus a _correlations/ layer linking people ↔ deals ↔ teams across them.

Day-one agent context

Ships with in-brain instructions — point Claude Code or GBrain at it and it knows the schema immediately.

Works with GBrain

Bootstrap your GBrain Company Brain from data you already have.

GBrain is the brain layer — search, synthesis, team serving. We're the importer that fills it on day one, from your fragmented exports, identity-resolved and privacy-clean.

just ask — company → gbrain
> "Build our Company Brain from company-data/ and emit the GBrain format." gbrain import brain/ # your Company Brain, populated

Not affiliated with the GBrain project. We emit the open-source GBrain format — we're the importer; GBrain is the brain.

Privacy-first

Your team's data never has to leave your machines.

The whole import runs locally before anything reaches a server or a shared brain — you stay the data controller throughout.

Employee PII never written

Emails and phone numbers are stripped in code at parse time — they never reach disk.

Bodies never read

Slack and mail content is never parsed. Only a frequency and recency signal survives.

HR · payroll · security quarantined

Sensitive exports are skipped by a hard-coded quarantine list — no adapter can bypass it.

You stay the data controller

Everything runs on your infrastructure before anything reaches a server. MIT-licensed — your security team can read every line.

sources/common.py · Collector
# Privacy enforced in code — no adapter bypasses it. class Collector: def add_person(self, name, **f): if EMAIL_RE.match(name) or f.get("phone"): return # employee PII never written self._merge(nk(name), f) QUARANTINE = {"HR", "Payroll", "Security", "messages"}

Local-first by construction. → Read the privacy model

What your brain can do

Give your company's brain a goal.

acme-brain — ObsidianClaudian
Company Brain · reads your vault
Pick a goal above to ask your company brain…

Get started

Bootstrap your Company Brain.

just ask — company → gbrain
# tell your agent (Claude Code / Codex, skill installed): > "Build our Company Brain from company-data/ and emit the GBrain format." gbrain import brain/

Full docs →

Prefer we do it?

Hand us your exports. Get back a populated brain.

We can onboard your company's brain for you — import, resolve, privacy-clean, and hand it back in GBrain or Obsidian. (Early access.)

Local-first · MIT · you're the data controller · emits to GBrain.