PIN
Personal Intelligence Network
A LITTLE NOTE, A NEW POSSIBILITY
Capture your thoughts. Connect the possibilities.
Find your next idea in the connections between your notes.
WELCOME
Where should PIN keep your notes?
PIN creates a pin_note folder with your first constellation, 01_Aquarius. Drop older PIN notes (brain-data.json or another PIN JSON file) into that folder and PIN reads them automatically. To move everything later, copy the folder.
QUICK CAPTURE
Capture a thought
LOCAL_ONLY excludes the note from cloud AI. For device-only storage, choose a local folder.
Your growing collection
Your notes and files are stored in your selected location.
To load previous notes, choose the folder containing brain-data.json in Settings → Storage. Existing notes are preserved.
DREAMING
Today's dream
MIND NODE
Your idea map
Your idea notes, grouped by project. Select a node to rename it, add ideas under it or open its note. Built from your notes without AI; nothing changes until you save.
Recent thoughts
View all notes →Your Library
IN THIS CONSTELLATION
Files in this folder
Files you put into this constellation folder or its subfolders. Text, Google Docs, Word, Excel, PowerPoint, Hangul (HWP) and PDF files become searchable in Search My Brain automatically. Images and scanned PDFs become searchable after Read text (cloud AI rules apply). The search box above filters this list too.
Thought Graph
Explore your connected mind.
Saved connections are solid; shared tags can appear as dashed suggestions. The star field and its faint strands are decoration only.
Zoom in to see keywords of the notes in view. Right-drag turns the brain at any zoom; when zoomed in, drag to move.
Mind Node
Shape your ideas and grow their connections.
Your brainstorming boards. Click a board to open it.
Mind Node
Double-click to add a card · drag + onto a card to connect · click a line to delete it · cards stay on the board until “Save as note”.
Keys: Enter edit · Tab new connected card · Delete remove · Ctrl+Z undo · Ctrl+Shift+Z redo · arrows move the view · Esc leaves Focus.
Ask Constellar
Ask across your notes, files, and connected ideas.
My data first, AI second. PIN searches your recent (hot) memory first and widens to older memory only when needed. The Coordinator picks only the agents a question needs: finding and summarizing stay on local AI; analysis and new ideas use your API key. Answers separate your notes, AI insight and web research, and every source links back to its note.
Recent AI runs on this device (0)
Kept only in this browser: the plan, the notes used, the model and how long it took. Nothing here is sent anywhere.
EXPLORE YOUR KNOWLEDGE
Search My Brain
Search your memory and research with evidence.
Your notes and cached file text first, and the public web only when you choose it. Every source is shown with its evidence.
Only this query is sent to web search. Leave private details out. Web search requires cloud permission and may incur provider charges.
Unread attachments are not downloaded. Cloud AI excludes LOCAL_ONLY notes and their shared attachments. Deep and Research add a model review, not independent fact checking.
Research notebook
The latest 30 reports are saved in your selected storage.
ARCHIVE SEARCH · OUTSIDE PIN_NOTE
Archive search
Search older files you keep elsewhere. Choose a folder once. PIN reads text, Word, Excel, PowerPoint, Hangul (HWP/HWPX) and PDF files, shows them here, and sends nothing to AI unless you select files and ask.
Advanced: skip a folder
Skip a folder inside a selected one (for example private papers).
CONSTELLAR · EVERY MORNING AFTER 9 AM (VANCOUVER)
Dreaming
Discover new connections in your memory.
Daily reviews run after 9 AM Vancouver when PIN opens. Weekly and monthly reviews are manual, bounded samples of recent and older notes. Closing PIN stops automatic runs. Accepting a suggestion saves its connection and idea to your editable map. API keys are tried first, then local AI. LOCAL_ONLY notes never go to cloud AI.
Past dreams
Support
Found a bug, have an idea, or just want to say hi? Send feedback by email — it opens in your mail app or in Gmail. Nothing is sent from PIN itself.
Tip: say what you were doing and add a screenshot if something went wrong.
Enjoying PIN?
PIN is free and made by one person. If it helps you think, you can buy me a coffee.
PREFERENCES
Settings
Storage
A local folder stays on this device and does not sync to your other devices. Your browser may ask for permission again. Site login and Drive connection are separate. You can connect a different Google account. New storage is a pin_note folder with numbered constellations (01_Aquarius, 02_Pisces …); existing PIN data stays untouched. Import rejects JSON with attachments.
CONSTELLATION
Drop PIN JSON files into this folder and PIN adds their notes when it opens. Other files you add here (subfolders too) appear under Library → Files in this folder and become searchable in Search My Brain.
Connect Google Drive to choose an account, or choose a local folder.
Google Drive
Open the folder that holds your files and choose Use this folder. PIN reads every file in it (subfolders too, originals are never changed) and saves your notes inside it in pin_note/01_Aquarius. Opening an existing pin_note or constellation folder uses it as is.
Older setup options
Entire Drive lets you search and read existing files; new notes go to a MEMO app folder. "Use this folder" loads a folder that already has brain-data.json. Original files are not changed or automatically sent to AI.
BRING YOUR OWN KEY
AI Settings
Paste any supported API key. PIN recognizes the provider, checks the key with that provider only, and then saves it. Keys are encrypted in your account and never shown again. API usage is billed separately by the provider; a ChatGPT or Claude subscription does not include API credit.
OpenAI, Anthropic, Gemini, xAI and DeepSeek keys are supported.
Which cloud AI to use
Quick answers still try local AI first. Deep reasoning tries your API keys first.
Save the key first. Model availability depends on your API account. Enter a custom text/chat model ID if it is not listed.
LOCAL_ONLY notes are always excluded from cloud AI. PDF and image reading follows the same Auto/Custom choice: Claude, ChatGPT and Gemini read PDFs and images, Grok reads JPEG/PNG images, and DeepSeek is skipped.
AI usage
| Provider | Model | Calls | Input tokens | Output tokens | Est. cost |
|---|
ON THIS DEVICE · NO AI KEY
What PIN learned
PIN learns from how you use your notes: links you make or dismiss, notes you drag onto a board together, and notes you open one after another (a note read for 10 seconds or more counts; one closed within 3 seconds counts against). Suggestions you pick lower in a list count a little more, and old signals fade by half every 60 days. Only note IDs and numbers are kept, in this browser — never note text.
Notes you use together
Suggestions you accept
| Kind | Linked | Dismissed | Weight |
|---|
Account
Admin
Accounts
| Sign-in | Joined | Last sign-in | Status |
|---|
ON THIS COMPUTER
Local AI
Run a downloaded model with Ollama or LM Studio. Choose a runtime, install it, then check the connection and select your model. Local settings stay on this device for your account.
Easy setup (recommended)
One file does everything: it installs Ollama if needed, lets PIN connect to it, checks this computer's real memory and graphics card, and downloads the best model for it. When it finishes, PIN opens and selects the model automatically.
- Windows: open the downloaded
pin-local-ai-setup.cmd. If Windows shows "Windows protected your PC", choose More info → Run anyway. Keep the black window open until it says Done. - Keep this page open. When the browser asks to access devices on your local network, choose Allow.
- The model download takes a few minutes (about 2–20 GB, depending on your computer).
Mac instructions
Open Terminal and run:
bash ~/Downloads/pin-local-ai-setup.command
macOS blocks double-clicking downloaded scripts, so Terminal is the reliable way.
Or choose a model yourself
How to check memory and graphics
Windows: Settings → System → About shows installed RAM. Task Manager → Performance → GPU shows "Dedicated GPU memory".
Mac: Apple menu → About This Mac shows the chip (M1 or later is Apple Silicon) and memory.
1. Install Ollama
Download for WindowsDownload for macOSLinux installation
Download the official installer, then open it and complete installation yourself. A website cannot install desktop software automatically.
Allow PIN to connect to Ollama
On Windows, quit Ollama from the system tray. Open PowerShell and run:
Keep that terminal open. For the desktop app, set the user environment variable OLLAMA_ORIGINS to and restart Ollama. Keep its default loopback address.
macOS/Linux environment setup. Allow local-network access if your browser asks. Browser restrictions can prevent a website from connecting; PIN will show a connection error.
1. Install LM Studio
Download LM Studio for your computer
Download and run the official installer yourself. In LM Studio, download a model that fits your computer, load it, then start the local server in the Developer tab on port 1234. Enable CORS in the server settings, keep the server on localhost, and allow local-network access in your browser if prompted.
2. Connect & choose a model
3. Download an Ollama model
Ollama must be installed and running first. Model downloads can take several minutes and use several GB. Choose a model that fits your RAM. Remote/cloud Ollama models are excluded.