🤖

eVera

AI Virtual Assistant

Voice-first AI assistant with 24+ specialized agents, 183+ tools, 3D holographic avatar, multi-modal interaction, and 4-layer memory system.

Application LayerPython, TypeScriptActive Development

View on GitHub

Key Features

24+ Specialized Agents (Productivity, Health, Finance...)

183+ Tools Library

Voice Control — 3 modes, 19 languages

3D Holographic Avatar — Three.js, lip sync, emotions

4-Layer Memory System — FAISS-powered

Tier-Based LLM Routing — Ollama/OpenAI/Gemini

Privacy-First — Local processing with Ollama

Architecture

Client Apps (Electron, React Native, Web)
├── FastAPI Backend
│   ├── LangGraph Orchestrator
│   ├── 24+ Specialized Agents
│   └── 183+ Tools Library
├── 4-Layer Memory (FAISS)
├── Tier-Based LLM Router
├── Voice Pipeline (STT/TTS, 19 languages)
└── 3D Avatar Engine (Three.js)

Code Example

python

from evera.memory import MemoryManager

memory = MemoryManager(
    working_capacity=10,
    episodic_ttl_days=90,
    semantic_index="faiss",
    procedural_learning=True
)

memory.store_working("User asked about weather")
memory.store_semantic("Tokyo avg April temp: 15°C")
context = memory.recall("What's the weather?", top_k=5)

API Highlights

Function	Description
`MemoryManager()`	Initialize 4-layer memory system
`memory.store_working()`	Store in working memory
`memory.recall()`	Retrieve relevant context
`agent.execute()`	Execute specialized agent action

← Back to All Products