Most apps that offer AI-powered summaries work the same way: you hand over your content, it passes through their servers, gets processed by an AI model, and a summary comes back. Simple. Convenient. And completely opaque.
What happens to your data in between? Who can read it? How long is it stored? Is it used to train future models?
Most apps don't answer these questions clearly — because the answers aren't reassuring.
EurekaVoice takes a different approach.
What Is BYOK?
BYOK stands for Bring Your Own Key. Instead of routing your voice note content through EurekaVoice's servers to generate AI summaries, you connect your own API key from an AI provider (such as OpenAI or Anthropic).
When you request a summary:
- Your transcription is sent directly from your device to the AI provider
- Using your API key — not ours
- The response comes back directly to your device
- EurekaVoice never sees the content of that request
We are not in the loop. We cannot read your notes. We have nothing to store, and nothing to leak.
Why This Matters More Than You Think
Your voice notes are personal
Voice recordings capture something uniquely intimate — the way you think out loud. Meeting notes, personal reflections, business strategies, medical reminders, creative ideas. This is not the kind of content you want passing through a stranger's server.
"We don't store your data" is not enough
Many services claim they don't store your data. But even if that's true, your content still travels through their infrastructure. It can be logged for debugging. It can be intercepted. It can be subject to legal requests. It can be exposed in a breach.
With BYOK, none of that applies. Your content never touches our infrastructure at all.
You control the relationship with the AI provider
When you use your own API key, you have a direct relationship with the AI provider. You can review their privacy policy, manage your usage, rotate or revoke your key at any time, and know exactly what terms govern your data.
What EurekaVoice Never Does
To be explicit about what our BYOK model means in practice:
- We never receive the content of your voice notes during AI processing
- We never store your transcriptions on our servers
- We never log your API requests or responses
- We never use your content to train or improve any model
- We never sell or share your data with third parties
The only data that leaves your device is what you explicitly send to the AI provider — using your key, under your terms.
How to Set It Up
Getting started with AI summaries in EurekaVoice takes about two minutes:
- Get an API key from your preferred AI provider (OpenAI, Anthropic, or others)
- Open EurekaVoice Settings → AI Summary
- Paste your API key — it's stored securely in your device's Keychain, never uploaded
- Record and transcribe as usual — summaries are now available on demand
Your key stays on your device. Your notes stay on your device. The AI provider processes only what you send, under your account.
The Trade-Off Worth Making
BYOK does mean one extra step compared to a "just works" AI feature. You need to create an account with an AI provider and manage an API key.
We think that trade-off is worth it — and we think you will too, once you consider what you're protecting.
Your voice notes contain your real thoughts. Your meetings, your ideas, your plans. That content deserves a privacy model that's honest about how it works, not one that buries the details in a terms of service document.
Privacy Is a Design Choice
At EurekaVoice, privacy isn't a compliance checkbox. It's a design constraint we apply from the start. BYOK for AI summaries is one expression of that — we deliberately chose an architecture that keeps us out of the loop, even though it would have been easier to build a centralized service.
We believe the best AI tools are the ones that work for you, not ones that harvest your data as a side effect.