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

Use cases · what a brain is for

Don’t search your data. Give your brain a goal.

A second brain built from your own data exports isn’t a filing cabinet — it’s a personal AI assistant with memory, a personal CRM that fills itself, a mirror that shows you what Facebook and Google know about you, and for teams, a company knowledge base your AI can use on day one. Every use case below names the exact prompt and the exact note in the vault that answers it — all of it 100% local, zero network calls, message text never read.

AI WITH MEMORYPERSONAL CRMTHE MIRRORGET HIRED / FUNDEDWHO KNOWS WHATDAY-ONE COMPANY BRAIN

Personal · every day

A personal AI assistant that actually knows you.

ChatGPT, Claude and Gemini all ship memory now — and every one of them keeps it in its own cloud, starting from zero. Your exports already contain years of who you know, where you’ve been and what you care about. Built into a vault, they become the AI memory you own: local, portable across models, full on day one.

00-me/identity.md

Your AI finally knows you

Read my brain, then answer like you've known me for ten years: what should I focus on this quarter?

Works with Claude Code, Codex & Obsidian
10-people/…

Never walk into anything blind

Brief me before my meeting with Sam: our full history across every platform, and three openers that aren't small talk.

One merged note per person, every source
Dashboard.md

Your network works for you

Who am I going quiet on that I shouldn't be? Show my dormant-but-strong relationships and draft a reconnect note in my voice.

The personal CRM you never have to maintain

Why nothing else does this: vendor memory starts empty and never leaves the vendor. A local AI assistant over your own vault starts full — from data the platforms are legally required to hand you.

Same model · different answer

Ask AI without a brain. Then ask it with yours.

The model isn’t the difference — the context is. Here is the same question, to the same AI, before and after it can read your vault.

any AI · no brain

Who should I reconnect with this month?

I don’t have access to information about your relationships or contact history. Could you tell me who your friends and colleagues are, when you last spoke, and which relationships matter most to you? Generally, experts suggest reaching out to people you haven’t spoken to in a while…

✕ no sources — it’s asking you to do the work
the same AI · your brain

Who should I reconnect with this month?

Three stand out from your dormant-but-strong list Dashboard.md: Dana K. — strength 4/5, last contact Nov 2023, you worked together 2019–2021 and she just changed companies 10-people/dana-k.md · Sam P. — 5 years of WhatsApp rhythm, quiet for 8 months 10-people/sam-p.md · Alex R. — met at the 2019 accelerator, invests in your space now 90-synthesis/network-map.md. Drafts in your voice are ready for all three.

grounded in: linkedinwhatsappfacebookgoogle takeout· 3 notes cited · message text never read

That’s the whole product. Same model, same question — the left one asks you to remember; the right one already knows, and shows its sources.

Personal · the mirror

See what the algorithms know about you.

Ever searched “what does Facebook know about me” or “what does Google know about me”? The answer is in the export they’re required to give you — ad interests, inferred traits, off-platform activity. The brain files it where you can finally read it.

buried in the export
facebook-export.zip
├── ads_information/
│   ├── advertisers_using_your_data…  712 companies
│   └── ad_preferences.json
├── other_logged_information/
└── apps_and_websites_off_of_facebook/
                               unreadable on purpose
your algorithmic mirror
vault/my-brain/50-mirror/
├── inferences.md              what they think you are
└── ad-profile.md              segments & advertisers

What are the strangest things the platforms believe about me — and what did they get wrong?

50-mirror/inferences.md · the demo people screenshot

Personal · the big moments

Get hired — or funded — through people you already know.

A warm introduction beats a cold application every time; the hard part is remembering who you know. The brain cross-references your applications, your network and your history into referral paths — the networking CRM for job seekers and founders that builds itself from your own data.

95-goals/job-search.md

Get hired

youDana · ex-colleague · strength 4hiring manager

"Companies I’ve applied to or searched for where I already know someone — ranked by tie strength, with a drafted note each."

Referral paths first, cold applications last.

95-goals/fundraising.md

Get funded

youAlex · 2019 acceleratorpartner at fund

"Map every path from my network to investors who back companies like mine, including second-degree intros."

Warm paths from data you already own.

And when it’s neither of those: 90-synthesis/network-map.md surfaces the strong relationships that went dormant — with a genuine, non-awkward reason to write to each one this week.

Personal · taste, recall, timeline

Your taste, your places, your life — one queryable graph.

Years of Spotify, YouTube, Maps, searches and check-ins add up to something no recommendation engine has: your actual taste and your actual history. Ask it anything — it’s your life timeline, and it answers.

95-goals/personalization.md

Recommendations that actually know you

Grounded in your saved places, reviews and years of listening — not in ads or what’s trending.

Somewhere to eat Friday I’ll actually like — from my saved places and reviews, not from ads.

80-search/search-log.md

Where did I see that?

Find that thing I searched or saved about sleep and training — around when I started running.

85-places/

Your life, mapped

Reconstruct my 2019: where I lived, traveled, what I listened to, who I talked to most.

85-places/ · 10-people/

Travel like a local — the local is you

The itinerary only your data can write: your saved spots, your people in the city, and the gaps worth filling.

Plan Rome from my own history: places I saved, people I know there, plus gaps worth filling.

Company · the Company Brain

The company knowledge base that starts full — from exports you already have.

Every AI knowledge management tool indexes live systems and assumes clean context exists. Your company’s real institutional knowledge is in the archives: Slack exports, Workspace directories, CRM dumps, decommissioned wikis, mail headers. One local build turns them into a brain — people, teams, deals and expertise, identity-resolved. Message text is never read.

01

Who knows this? Who owns this?

Who knows payments? Best person, their team and channel, one backup — and how warm my path to them is.

From roles, channels & authorship — not surveillance
02

Memory that survives turnover

Maria resigned this morning. Which accounts and deals did she own, who did she talk to most, and what should her successor do in week one?

Knowledge transfer when an employee leaves
03

Day one, fully briefed

Draft my week-one onboarding plan: the 5 people to meet with why, the 3 channels to read first, the questions to ask each team.

Works for a new hire — or a new AI agent
95%

of enterprise GenAI pilots deliver no measurable P&L impact — MIT names the root cause the “learning gap”: AI with no memory of the business. Fill the brain first, from data you already own, and every agent you deploy inherits it. The full company story →

The honest limits

What it won’t do — said upfront.

Freshness

It’s a snapshot, not a live feed

Sharp on history, blind to this week. Re-export and update in place with --refresh — your edits are kept, and the run is summarized in _UPDATE_REPORT.md.

Privacy by design

Message text is never read

WhatsApp, Slack, email, tickets: relationship signal is who, how often, how recent — never the content. That’s the code, not a policy.

Coverage

Thin where you were thin

The brain is as complete as the data you feed it. Barely used a platform? That slice will be thin — and it will say so.

FAQ

Questions people actually ask.

What is a second brain app, and how is this one different?

A second brain app is an external, searchable system for what you know — notes, people, ideas. Every second brain and personal knowledge management tool starts empty and asks you to feed it. Second Brain Link starts yours full: it builds the vault from years of your own data exports (LinkedIn, Google Takeout, Instagram, Facebook and 24 sources in all) in about a minute, locally.

How do I see what Facebook or Google actually knows about me?

Request your official export (Facebook: Accounts Center → Download your information; Google: takeout.google.com). Both include the advertising and inference data platforms hold on you. Second Brain Link turns those files into a readable "algorithmic mirror" — your ad interests, inferred traits, and off-platform activity — as plain notes on your own machine.

Can my AI assistant have memory of my life without sending my data to a cloud?

Yes. ChatGPT, Claude and Gemini all ship memory now, but each vendor’s memory starts empty and lives on their servers. Second Brain Link builds a local, portable memory layer from your own exports — plain Markdown any AI can read. The transform runs entirely on your machine with zero network calls, and message text is never read.

What is a Company Brain?

A structured, AI-readable knowledge base of your company: people, teams, accounts, deals, wiki knowledge and collaboration signal, resolved into one graph. Second Brain Link bootstraps it from exports your company already has — Slack, Google Workspace, LinkedIn Company, CRM, wikis, mail archives — so agents and new hires inherit institutional memory on day one.

How is this different from enterprise search tools like Glean or Microsoft 365 Copilot?

Those tools index live, connected systems in the cloud — and they assume clean context already exists. Second Brain Link fills the brain first, from the historical export archives they skip (including the context of people who already left), and it runs locally: your company’s brain never leaves your machines. The two approaches compose — we’re the cold-start layer, not another index.

Is my data uploaded anywhere?

No. The core transform makes zero network calls — verifiable in the open MIT-licensed source. Message text, email bodies and ticket prose are never read (relationship signal is who/how-often/how-recent only), and there is no account and no telemetry. Cloud sync exists only as an optional paid product.

What should I do with my LinkedIn data export or Google Takeout?

Don’t let it rot in your downloads folder. Drop the zip on Second Brain Link — CLI, Studio, or the Agent Skill in Claude Code / Codex — and it becomes a connected, private knowledge graph: your people, career, places, interests and voice, ready for any AI to reason over. See the Sources page for every supported export.

Your data already exists. It’s already yours.

Free & open source (MIT) · 100% local · from zips to a working brain in about a minute.