Core Services
The practical map of Viventium's chat, voice, Telegram, scheduling, memory, Prompt Workbench, workspaces, and workflow services.
The Product Map
Viventium is a brain-inspired cognitive system, but the easiest way to understand it is to look at the services you actually use.
Each service has a plain user job and a technical backbone. The goal is not to collect features for a checklist. The goal is one system that can think with you, remember useful context, speak with you, work between conversations, and let you inspect what is happening.
| Service | What you get | Where to read next |
|---|---|---|
| Chat Web App | The main desktop conversation surface for long-form thinking, connected accounts, files, tools, agents, and background follow-through. | Voice, Chat & Messaging |
| Real-Time Voice | A spoken surface for thinking out loud, interruptions, Wing Mode, Listen-Only Mode, and provider-aware speech output. | Voice, Chat & Messaging |
| Telegram + Mobile Messaging | Mobile continuity with text, voice notes, voice replies, reminders, briefings, and worker callbacks away from the browser. | Voice, Chat & Messaging |
| Scheduling Cortex | Morning briefings, reminders, recurring research loops, and proactive updates that stay quiet when there is nothing useful to add. | Scheduling |
| Memory + Recall | Durable facts, recent working context, transcripts, project state, and selective retrieval instead of one stale prompt blob. | Memory & Continuity |
| Prompt Workbench | A local control room for source, live, and evaluated prompts, with drift review, evals, prompt traces, and scheduled prompt work. | Prompt Workbench |
| GlassHive Workspaces + Workers | Persistent computer-like workspaces where AI workers can use a browser, files, and terminal while you watch, steer, pause, or take over. | Projects, Workers & GlassHive |
| Self-Healing, Feature Requests, Bug Reports | AI-assisted local workflows for diagnosis, specifications, repro steps, RCA, proposed fixes, approval gates, and QA evidence. | Self-Healing & Feature Requests |
| Connected Tools, Browser, Analysis | Connected accounts, web search, crawling, file analysis, research, and artifact generation when chat needs live context. | Connected Workspaces |
The Main Services
Chat Web App
The chat web app is the main place you sit with Viventium. It is where longer reasoning, visible drafts, connected accounts, uploaded files, tool calls, agent selection, and background follow-through come together.
For builders: the current product stack uses Viventium's own web-chat fork as the primary web app. That matters because the chat surface is not a thin demo UI. It owns the day-to-day conversation experience and connects to the same memory, scheduling, voice, Telegram, and worker layers.
Voice Gateway
Voice is for the moments where typing slows thinking down. Viventium uses a real-time voice surface and a Voice Gateway so spoken turns can use the same assistant pipeline as chat instead of becoming a separate toy mode.
That means voice can share continuity with chat, support natural interruption, expose Wing Mode and Listen-Only Mode, and clean up text before speech so the assistant does not read raw URLs, citations, code fences, or provider markup at you.
For builders: voice can be fully local when both speech-to-text and text-to-speech use local routes. Hosted speech providers are explicit choices; when selected, the relevant audio or text goes to that provider.
Telegram Bridge
Telegram keeps Viventium useful when you are not in the browser. You can use text, voice notes, voice replies, scheduled briefings, reminders, and mobile follow-up without treating the phone as a separate assistant with no memory.
The important point is continuity: Telegram should feel like the same Viventium, not a weaker bot bolted onto the side.
Scheduling Cortex
Scheduling is how Viventium helps between conversations. It can run reminders, morning briefings, recurring research, and project momentum checks, then deliver the result to chat or Telegram.
The design rule is simple: proactive AI should be useful or quiet. If a scheduled run has nothing meaningful to add, silence is better than filler.
Memory + Recall
Memory is not just "more chat history." Viventium separates durable facts, recent working context, signals, drafts, transcripts, and project state so future work starts sharper without flooding every prompt with stale context.
That is what lets chat, voice, Telegram, schedules, and workers feel like parts of one system instead of separate apps with separate memories.
Prompt Workbench
Prompt Workbench is the local QA and control surface for the brain of the product. It shows prompt state across source, live, and evaluated forms, then gives builders a safer way to review drift, draft changes, run evals, inspect traces, and understand scheduled prompt behavior.
This is part of the reliability story: Viventium does not treat prompts as invisible magic strings. It gives them a workbench.
GlassHive Workspaces + Workers
GlassHive is the persistent worker runtime behind longer-running work. A workspace is a computer-like environment where an AI worker can use a browser, files, terminal, and tools while you can watch what it is doing, steer it, pause it, resume it, take over, or inspect artifacts.
That is the difference between "the AI says it did something" and "you can see the work being done."
Self-Healing, Feature Requests, Bug Reports
Viventium also has local workflow surfaces for improving and repairing the system. Self-healing starts with diagnosis and RCA. Feature requests are turned into success criteria and QA acceptance before implementation. Bug reports collect reproduction details and proposed fixes before code changes.
The safety detail matters: these workflows are explicit operator actions. They are not triggered by keyword-matching random chat messages, and they use approval gates before implementation.
Why This Fits The Brain-Inspired Model
The services above are not disconnected modules. They map to a cognitive system:
- Chat and voice are the conscious conversation surfaces.
- Background agents and Red Team are specialized supporting cortices.
- Memory and recall are continuity systems.
- Scheduling is prospective attention.
- GlassHive workers are the motor layer for real work.
- Prompt Workbench is the metacognitive inspection surface.
- Self-healing and feature-request workflows are the product learning how to improve itself with human approval.
That is the core idea: one coherent assistant in front, many specialized systems behind it, and enough visibility that people can actually trust the work.
Keep Reading
- Voice, Chat & Messaging - The user-facing surfaces in detail
- Prompt Workbench - The local prompt control and evaluation surface
- Self-Healing & Feature Requests - AI-assisted repair and improvement workflows
- Projects, Workers & GlassHive - Durable workspaces and worker execution