Trust Model & Deployment
How local-first control, connected accounts, cost discipline, and the current public install path fit together.
Why Deployment Matters
For Viventium, deployment is not just an ops detail.
It changes:
- who controls the runtime
- where data lives
- how much trust you have to give away
- how much you pay to access frontier models
That is why the public docs treat deployment as part of the product story.
The Three Big Trust Questions
| Concern | Viventium approach |
|---|---|
| Blind spots | One main assistant can pull in specialized background agents instead of relying on one narrow model response |
| Cost | Use the OpenAI or Anthropic account you already have, or bring your own API keys for other supported providers and models, instead of paying extra wrapper markup |
| Privacy and visibility | Local-first posture, open source code, and inspectable config instead of "just trust us" |
The Current Public Path
Today, the public-ready path is:
- Start from the GitHub repo
- Run the local installer
- Open
Settings -> Connected Accounts - Either connect your existing OpenAI or Anthropic account, or configure API keys for another supported provider and model
The current setup also expects a Groq API key for the fast wake-up checks that help decide whether background agents should activate.
What Runs Where
| Component | Where it runs | Why |
|---|---|---|
| App surface, config, runtime state | Your machine | Full control, inspectable |
| Voice processing (Whisper C++, Chatterbox) | Your machine | Audio never leaves your computer |
| Conversation history and memory | Your machine | Private by default |
| Main AI inference (GPT, Claude, and others) | Your connected account or your own API keys | No markup, and you choose the provider/model path you want |
| Activation detection | Groq | Fast, cheap wake-up checks for background agents |
| Background agent execution | Your connected account or your own API keys | Same model path, independent analysis |
| Web search (SearXNG) | Self-hosted container | Private search, no tracking |
The key principle: what can run locally, does. What needs a frontier model goes through the account you already own. Nothing goes through a Viventium middleman.
Why Open Source Matters Here
Open source matters because Viventium is meant to be trusted with:
- important thinking
- work context
- voice conversations
- project state
- future autonomous follow-through
When the system matters that much, visibility is part of the product value.
What This Means For Cost
The cost stance is simple:
- use the subscriptions you already have where possible
- keep lightweight checks lightweight
- avoid making users pay markup on top of access they already purchased
That is a meaningful part of the value proposition, not a footnote.
What Is Still Evolving
The current public path is macOS-first local install. What is actively being developed:
- Broader platform support beyond macOS
- More polished packaging and installer
- Managed deployment paths for teams
- Broader GlassHive sandbox profiles
These are real directions, not promises. The docs stay honest about what is public and reliable today.
Keep Reading
- Install & Setup — Follow the current public flow
- Architecture Overview — See every component and where it runs
- Product Philosophy — Why trust and ownership matter so much here
- Troubleshooting — Recover quickly when the local path breaks