BYOAI

Tareea has no built-in AI — you plug in your own provider and API key, your browser talks to them directly, and every action is a click you deliberately took.

How it works

Notes and tasks are the product. If you want AI on top of them, you connect one of seven providers with an API key you already have. When you click an AI action, your browser sends the content directly to whichever provider you chose, using the key you saved — Tareea's servers aren't in that call. No proxying, no logging, no caching. When the answer comes back, it lands in a review panel where you decide whether to keep it.

Each provider ships with three model tiers — best, balanced, fast — editable in Settings. The consent screen names the provider and confirms the key never leaves your browser; provider-specific data-use terms live at the provider itself, so you can read them at the source before you connect.

What it can do

Eleven actions across notes, Folios, and tasks. Every result lands in a review panel — copy, save as note, insert, or discard. Nothing changes until you tell it to.

Private mode

This is architectural, not policy. Encryption that an AI provider could read isn't really encryption — so Private mode removes the AI code path entirely rather than just hiding a toggle.

Pro

BYOAI is a Pro feature. It runs in workspaces using Connected encryption mode — which is itself Pro-only, so from your perspective it's one gate, not two. Private (fully end-to-end-encrypted) workspaces have zero AI surface — not hidden behind a setting, but structurally absent. The credential-store table doesn't exist for a Private workspace, the AI menus don't render, and there's no code path for content to reach an external endpoint.

The honest tradeoff

Most AI tools run constantly in the background, processing and syncing whether you're using them or not. Tareea's AI is called per feature, per click, never ambiently. For most users — anyone who never connects a key — none of it runs at all.

How this is ethical, not just private

The same principles responsible AI design is generally held to, applied here as architecture rather than policy.

Privacy by design

AI requests route directly from your browser to the provider you chose — never through us. We don't see the prompt, the response, or store anything about the exchange. This isn't a promise to handle your data carefully. It's a structure where there's nothing here to handle in the first place.

Human autonomy

Private mode doesn't limit AI — it removes it entirely, on request, with nothing lost elsewhere in the product. Respecting a person's right to decline a technology means declining costs them nothing. That's the bar Tareea is built to meet.

Safety by restriction

Your browser can only talk to providers you've explicitly connected — nothing else, by design, enforced at the browser level, not by a policy we're asking you to trust. A system is safer when the harmful path isn't just discouraged, it's unavailable.

Fair access

Pro costs the same whether you use BYOAI or not — there's no separate charge for connecting a provider, no AI tier priced above the base plan. The only cost tied to AI itself is whatever your chosen provider bills you directly, on the terms they publish.

What this can't reach

Once a request leaves your browser for the provider you chose, what happens to it is between you and that provider. We don't control, and can't vouch for, the bias or fairness of what a model outputs, how explainable its reasoning is, or the environmental cost of running it. Those are real, and they belong to the provider you picked, on the terms they publish — not to us.

If that matters to you, it's worth reading the provider's own policy before connecting a key, the same way you'd want to know it about us.