Your Data — Data Sovereignty
The most valuable thing you produce as a knowledge worker is context.
SoY was designed so you can't give it to us — even if you wanted to.
In December 2025, Meta acquired Limitless AI. Limitless was an AI wearable that captured everything you said throughout your day — your meetings, your conversations, your relationships. All of it, centralized in their cloud.
Then Meta owned it.
This is not a warning about Limitless. It's proof of a design problem. When your data lives on someone else's server, your context becomes an acquisition asset — not yours.
Acquired by Meta · December 2025
The Data Centralization Problem
2023
Limitless ships its AI wearable
The product promise: capture everything — your meetings, your phone calls, your conversations — and let AI surface insights. Useful. The data lived on Limitless's servers.
2024–2025
Users build context over time
People wore the device. Built up months of captured context. Their professional relationships, client conversations, personal discussions — all of it accumulating in Limitless's cloud.
December 2025
Meta acquires Limitless AI
The users' context — everything they'd said and heard for two years — was now owned by Meta. Not because they chose to share it with Meta. Because the architecture made it possible.
The structural problem
"Central storage creates acquisition value. Acquisition value creates acquisition risk."
This isn't a story about one company making a bad deal. It's a story about a design pattern. Any tool that stores your data centrally has created something that can be bought, sold, breached, or shut down.
The solution isn't better terms of service. It's architecture that makes centralization structurally impossible. SoY has no servers to acquire. There is nothing to breach. Your context was never there.
Architecture
Zero cloud. Zero servers. Zero sync. Every module SoY tracks — your contacts, clients, projects, decisions — writes directly to a SQLite database on your local filesystem. Nothing else.
Every other tool
Your machine
Where you enter the data.
Their cloud
Where your data actually lives. Their servers. Their infrastructure. Their rules.
Their model training
Your data as product. Check the ToS.
Their acquisition value
When they sell, your context sells with them.
Data location
https://theirapp.com/api/your-data/...
SoY
Your machine
Where you enter the data.
Your machine
Where your data actually lives. SQLite file. Your filesystem. Your control.
Full stop.
That's the entire chain. Nothing else is in it.
Data location
~/.local/share/software-of-you/soy.db
Local-First
Not a marketing claim. Five concrete truths about what it means for your data to live on your machine.
01 / 05
If SoY shuts down tomorrow
Your data doesn't.
It's already a SQLite file on your machine. SoY can cease to exist and your context continues to exist, intact, on your filesystem — readable by any SQLite client.
02 / 05
If you cancel and want to export
There's nothing to export.
You already own the file. There's no export button because there's no centralized data to extract. The file is already yours. It was always yours.
03 / 05
If your internet goes down
SoY works.
It doesn't know the internet exists. There's no API call to fail, no sync to interrupt, no cloud dependency to lose. Your context is on your machine. That's where it runs.
04 / 05
If you want to delete everything
rm ~/.local/share/software-of-you/soy.db
Done. No deletion request, no support ticket, no 30-day hold period. One command. It's gone because it was yours to remove.
05 / 05
What SoY has seen of your data
Nothing. It was never designed to.
SoY has no telemetry pipeline pointed at your database. No sync process that uploads context to our servers. No analytics that phone home with your records. This is not a privacy policy promise — it's a structural fact. We built no mechanism to receive it.
The Technical Case
Claude Code runs locally — or via an API key you control. SoY passes context directly from your local database to your local Claude session. The data path is complete.
🗄
Source
soy.db
SQLite database on your local filesystem. Your contacts, clients, projects, decisions — structured and cross-referenced.
⚡
SoY Plugin
Context injection
SoY reads the local database and injects structured context directly into your Claude Code session. No intermediary. No upload.
🤖
Destination
Your Claude session
Claude receives your context. Locally, through the API key you already control. The model processes it and returns to you.
✓
What leaves your machine
Your Claude API call
The same API call you'd make anyway. SoY adds context to it. Nothing more is transmitted.
No middleware. No SoY servers in the path. No third party sitting between your database and your model. The entire chain: your machine → your Claude. SoY is a plugin, not a platform. It was designed to stay out of the way.
The Design Paradox
The most useful AI needs the most context. Most tools solve this by centralizing your data. We solved it by never having it.
There's a design assumption in AI tooling: to give you intelligence, we need your data on our servers. So the model can be trained on it. So we can sync it. So we can analyze it across users.
SoY proves that assumption wrong. The context you need isn't on our servers — it's on your machine. We connect your local database to your local Claude. The intelligence comes from structure and locality, not from centralization.
You shouldn't have to trade privacy for intelligence. You don't.
How It Works vs. How They Work
How most AI tools work
Your data → their cloud
Your data enters the tool
Stored on their servers, in their database, under their terms
Processed by their model (trained on what, exactly?)
Insights returned to you
Their ad targeting. Their model training. Their acquisition value. Check the ToS.
How SoY works
Your data → your machine
Your data enters SoY
Stored on your machine, in your SQLite file, under your control
Passed to your Claude session via API key you already own
Insights returned to you
Full stop. Nothing else is in this chain.
Google User Data Disclosure
Software of You optionally connects to your Google account to sync Gmail, Google Calendar, and Google Docs data. Here's exactly what happens with that data.
Data Accessed
What Google data does SoY access?
Data Usage
How does SoY use your Google data?
All Google data is stored locally on your machine in a SQLite database file. SoY uses this data to:
Your Google data is processed by Claude (Anthropic's AI model) only within your local Claude Code session, using your own API key or subscription. SoY does not use your Google data for model training, advertising, analytics, or any purpose other than providing you with the features described above.
Data Sharing
Is your Google data shared with anyone?
No. SoY does not share your Google user data with any third parties. Specifically:
Data Storage & Protection
Where is your Google data stored and how is it protected?
~/.local/share/software-of-you/soy.db~/.local/share/software-of-you/google_token.json, protected by filesystem permissions~/.local/share/software-of-you/backups/Data Retention & Deletion
How long is your data kept, and how do you delete it?
Retention: Your Google data is retained on your machine indefinitely — until you choose to delete it. SoY does not automatically expire or purge your data.
Deletion: Because all data is stored locally, you have full and immediate control over deletion:
rm ~/.local/share/software-of-you/soy.dbrm ~/.local/share/software-of-you/google_token.json — or revoke access from your Google Account permissions pagerm -rf ~/.local/share/software-of-you/No deletion request, no support ticket, no waiting period. Your data is a file on your machine — deleting it is instantaneous and complete.
Software of You's use and transfer of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements.
Built Differently
You shouldn't have to choose. With SoY, you don't.
"You are not talking to AI. You are thinking with it."