How AI Works in Clairvoyant
One of the first questions privacy-conscious users ask is: "If Clairvoyant has AI features, does that mean my data gets sent to some AI company?"
No. Here's exactly how it works.
Most of It Isn't Even AI
About 85% of Clairvoyant's "smart" behaviour is actually deterministic pattern matching — rules-based logic that runs instantly on your device. When you type something like "buy groceries tomorrow at 5pm," the app doesn't need a neural network to figure out what you mean. It parses the input locally, on your device, in every mode.
Only the remaining 15% of inputs — genuinely ambiguous or complex requests — ever need an actual AI model. And even then, you control where that processing happens.
The Default: Zero-Knowledge
Out of the box, Clairvoyant is zero-knowledge. The server stores your synced data as encrypted blobs it cannot read. AI features that need a model will try to run on your device if you have a local AI setup (like Ollama with open-source models). If you don't, those features simply aren't available until you either set up local AI or opt in to Server-Assisted mode.
The default gives you the strongest possible privacy guarantee. There is no server to trust, no connection to monitor, no policy to read. The math simply doesn't allow anyone else to access your data.
Optional: Server-Assisted AI
If you want more powerful AI features without running models on your own hardware, you can opt in to Server-Assisted AI. Here's what that means and what the trade-offs are:
What you get:
- AI features work on any device, including phones and low-spec hardware
- No setup required — it just works
What it means for your privacy:
- Our server temporarily decrypts your tasks in memory only to run AI, then immediately discards the plaintext
- Your data is never stored in plaintext, never written to logs, never used to train models
- We run our own AI models on our own servers — no third parties involved
- You're trusting Clairvoyant's server to handle your data responsibly during processing
What you can always do:
- Turn it off at any time from Settings and return to full zero-knowledge
- Your encryption keys remain yours — revoking access means the server can no longer decrypt anything
This is a genuine trade-off. You go from "the server mathematically cannot read my data" to "the server processes my data in memory but doesn't store it." We think that's a fair trade for powerful AI features on any device — but we want you to make that choice knowingly, not have it made for you.
No Third-Party AI Companies. Ever.
This is worth emphasising: in all three modes, no third-party AI company ever sees your data. Not OpenAI, not Google, not Anthropic, not anyone else.
Many productivity apps that advertise "AI features" are really just forwarding your tasks, notes, and calendar data to a third-party API. Your data lands on someone else's servers, governed by someone else's privacy policy, potentially used to train someone else's models.
Clairvoyant doesn't work that way. By default, AI runs on your device (or doesn't run at all if you haven't set it up). If you opt in to Server-Assisted mode, it runs on our own infrastructure. There is no third party in the chain, ever.
What Always Stays on Your Device
Regardless of which AI mode you choose — even Server-Assisted — certain data never leaves your device:
- Productivity patterns — when you're most focused, how long you work, your energy levels throughout the day
- Work habits — how you use the app, what features you rely on, your routines
- Behavioural data — everything the app learns about how you work best
This data is used locally to personalise your experience (like suggesting optimal times for deep work), but it is never transmitted to any server in any mode.
Your Keys, Your Data
By default, your encryption keys never leave your device. The server literally cannot decrypt your data — it only ever sees encrypted blobs.
If you opt in to Server-Assisted AI, you're granting temporary, revocable access for AI processing. The moment processing is complete, your decrypted data is discarded from memory. Nothing is persisted, nothing is logged.
You can disable Server-Assisted AI at any time from Settings. Doing so immediately cuts off all server-side processing and returns you to full zero-knowledge.
How This Compares
| Clairvoyant | Most Competitors | |
|---|---|---|
| Where AI runs | Your device (default) or our servers (opt-in) | Third-party AI company |
| Who sees your data | Only you (default) or temporarily us (if you opt in) | The app company, their AI provider, potentially others |
| Data used for training | Never | Often yes, buried in terms of service |
| You can audit it | Yes — open source | No |
| Behavioural data | Always on-device | Usually sent to servers |
Still Have Questions?
If something here isn't clear, or you want more technical detail, reach out:
- GitHub Issues — for technical questions about the implementation
- Email — support@clairvoyant.one