What is Viventium
A brain-inspired, local-first AI system that brings chat, voice, Telegram, memory, scheduling, workers, and prompt governance into one cognitive product.
The Short Version
Viventium is a brain-inspired AI system, not just another chatbot.
The main assistant stays coherent in front. Behind it, specialized systems handle memory, voice, Telegram, scheduling, connected tools, background agents, worker environments, prompt review, and AI-assisted repair workflows.
That matters because real decisions and real work need more than one model in one chat box.
Start from the public
GitHub repo,
have the required background-agent activation key ready, run the local
installer, then use Settings -> Connected Accounts or your own
provider keys for the model path you want.
What You Actually Get
Viventium combines the core services a useful cognitive system needs:
- Chat Web App - the main desktop surface for reasoning, files, tools, agents, connected accounts, and background follow-through
- Real-Time Voice - spoken conversation with interruption, provider choice, Wing Mode, Listen-Only Mode, and speech-safe output
- Telegram + Mobile Messaging - mobile continuity with text, voice notes, voice replies, reminders, scheduled briefings, and worker callbacks
- Scheduling Cortex - morning briefings, reminders, recurring research loops, and proactive updates that stay quiet when there is nothing useful to add
- Memory + Recall - durable facts, recent context, transcripts, drafts, and project state without stuffing every prompt with stale history
- Prompt Workbench - a local control room for source/live/evaluated prompts, drift review, evals, traces, and scheduled prompt work
- GlassHive Workspaces + Workers - persistent computer-like workspaces where AI workers can use browser, files, and terminal while you watch or take over
- Self-Healing, Feature Requests, Bug Reports - explicit AI-assisted workflows for RCA, specs, approval gates, isolated worktrees, and QA evidence
- Connected Tools, Browser, Analysis - connected accounts, web search, crawling, files, research, and artifact generation
Start With The Right Docs
Core Services
The practical map of the chat, voice, Telegram, scheduling, memory, workers, Workbench, and workflow surfaces.
Voice, Chat & Messaging
The chat web app, real-time voice, Wing Mode, Listen-Only Mode, Telegram text, and Telegram voice.
Prompt Workbench
The local prompt QA surface for drift review, evals, traces, scheduled prompts, and safer prompt changes.
Self-Healing & Feature Requests
How local AI workflows turn bugs and ideas into RCA, specs, approval gates, and QA evidence.
Why People Reach For It
Viventium is built around four core problems:
-
Single-AI blind spots - If you talk to one AI with one system prompt and one model, you risk confirmation bias and missing perspectives. Viventium can activate specialized background agents when the conversation needs deeper checking, broader research, or independent challenge.
-
Cost and provider choice - Third-party wrappers add markup and lock-in. Viventium can use your connected account path or your own API keys across supported providers and models. Local voice routes can reduce voice cost when you choose them.
-
Local-first control - The runtime, config, memory/control plane, artifacts, and workflow state live locally. When you choose hosted model, speech, or tool providers, those calls go to the provider you selected instead of being hidden behind a black box.
-
Agentic autonomy with visibility - Chat alone cannot do real work. GlassHive gives AI workers persistent workspaces you can watch, steer, pause, resume, or take over.
What The Docs Cover
Start here
Foundations
- Product Philosophy
- Architecture Overview
- Brain-Inspired System
- Background Agents
- Red Team Cortex
- Memory & Continuity
Surfaces and workflows
- Voice, Chat & Messaging
- Prompt Workbench
- Self-Healing & Feature Requests
- Connected Workspaces
- Scheduling
- Browser & Analysis
- Projects, Workers & GlassHive
What Makes Viventium Different From A Chatbot
| Typical chatbot | Viventium |
|---|---|
| One model, one prompt | Main assistant plus specialized background cortices |
| No independent checking | Red Team cortex challenges weak assumptions and unsupported claims |
| Locked to one provider | Connected accounts and provider/model choice |
| Chat only | Chat, voice, Telegram, scheduling, memory, workers, and prompt governance |
| No durable work environment | GlassHive workspaces let workers use browser, files, and terminal with user visibility |
| Prompt changes are invisible | Prompt Workbench exposes source, live, evaluated, drift, evals, and traces |
| Bug reports become vague tickets | Local workflows collect RCA, repro, specs, approval gates, and QA evidence |
Where To Go Next
- Core Services - See the full product map
- Install & Setup - Follow the current public-ready install flow
- Voice, Chat & Messaging - Understand the user-facing surfaces
- Product Philosophy - Read the deeper "why"