For teams already on Google Cloud
Bring your own Vertex AI key.
The memory stays on your machine.
Mnemosyne OS already speaks to Gemini through Vertex AI. Add your service account in Settings, and calls run under your own Google Cloud project, on your own invoice. Your vaults, your recall and your writing stay on your machine either way.
How it connects
One key, in Settings.
Open Settings → AI & Models, add your Vertex AI service account key (a JSON file), and pick it for reasoning or chat. Mnemosyne OS routes those calls straight to Google Cloud. Pair it with a local model, or with any other provider Mnemosyne OS supports, and switch per task.
Region, audit logs, a dedicated deployment
Have a checklist? Send it over.
Some organizations need to pin a region, review Cloud audit logs, or scope a deployment to their own project before anyone touches it. Send us your real list instead of letting us guess. We will go through it with you.
Questions
What people ask before wiring Vertex up.
- How do I connect Vertex AI to a local-first second brain?
- Open Settings → AI & Models in Mnemosyne OS, add your Vertex AI service account JSON file, then pick it for chat or reasoning. From there the calls go to your own Google Cloud project. The vaults, the index and the recall stay on your machine. How it works →
- Is there a sovereign alternative to a hosted AI assistant?
- Mnemosyne OS keeps the memory on your machine. You pick the engine task by task: a local model downloaded once, your own key with a provider, or your own server on your network. With Vertex AI the call runs through your Google Cloud project, on your invoice. You never open an account with us. Local-first, in detail →
- Do my documents go to Google?
- Your question and the retrieved passages reach the model, at the moment you ask. Indexing and search run on your machine. A vault marked MAXIMUM answers only questions that name it. Its content reaches a cloud model only if you confirm it, each time.
- Vertex AI or a plain Gemini API key: what changes for a team?
- Both routes ship. A plain API key hangs off a separate account. Vertex AI runs through your organization’s Google Cloud project, with its access control, its quotas and its billing. Some organizations already require that path.
- Does it work with no cloud at all?
- Yes, with a local model downloaded once. Indexing, search and the answer then run on the machine. The moment you pick a cloud model for a task, your question and the retrieved passages go to that provider. You make that choice per conversation.
- Can I pin a region or get a dedicated deployment?
- Not self-serve today. The calls already leave from your own Google Cloud project. Pinning a region, reviewing audit logs or scoping a dedicated deployment is something we go through with you before anyone starts. Send your list and we will read it. Write to us →
- Who bills, and how much?
- The download is the complete OS, free, with no account. With your Vertex key, Google bills your usage directly and we are not on that path. A local model costs nothing once it is downloaded. Download →
Your memory stays local.
Your Google bill stays yours.