Your AI.
Self-hosted.
Private.

Voice conversation, vision, document memory, and web search — running on your hardware. No cloud. No data leaves your machine.

curl -fsSL https://molebieai.com/install.sh | bash

Get the latest updates

What You Get

Three Thinking Modes

Instant answers, chain-of-thought reasoning, or extended deep thinking. Pick the depth that fits your question.

Voice Conversation

Wake-word activation, speech-to-text via Whisper, natural text-to-speech via Kokoro, with speaker verification.

Image Understanding

Drop, paste, or attach images. Your AI sees and understands visual content alongside text.

Document Memory

Upload PDFs, DOCX, TXT, and Markdown. Hybrid vector + BM25 search with cross-encoder reranking for accurate retrieval.

Web Search

Self-hosted SearXNG with LLM-powered intent classification. Get answers grounded in real-time web results with source citations.

Full Data Ownership

Everything in SQLite on your machine. Multi-user with full isolation. No telemetry, no cloud, no exceptions.

See It in Action

CLI tool — one-command install process

Three Thinking Modes

01

Instant

Fast, no chain-of-thought

Click to expand

Instant mode — sub-second responses

02

Thinking

Chain-of-thought reasoning

Click to expand

Thinking mode — 2048 token budget

03

Think Harder

Extended deep reasoning

Click to expand

Think harder — 8192 token budget

Key Features

Voice Conversation

Click to expand

Wake word, speech-to-text, natural TTS response

Web Search

Click to expand

Real-time search with source citations

Up and Running in 60 Seconds

01

Install

One command pulls the repo and sets up everything.

curl -fsSL https://molebieai.com/install.sh | bash
02

Auto-Configure

Detects your hardware, picks optimal models, generates config.

Apple Silicon detected → MLX backend → 16GB RAM → Balanced profile
03

Chat

Open localhost:3000 and start talking to your AI.

molebie-ai run

Why Self-Host?

Cloud AI

Molebie AI

Data Privacy

Your data on their servers

Everything stays on your machine

Monthly Cost

$20–100+/month per user

Free forever, you own the hardware

Offline Access

Requires internet connection

Works completely offline

Customization

Limited to what they offer

Full control over models and config

Vendor Lock-in

Switching costs and data export pain

Open source, standard formats

Compliance

Trust their compliance claims

Verify yourself — it’s your infra

Zero Telemetry

100% Offline Capable

MIT Licensed

SQLite — Your Data

No Cloud Dependencies

System Requirements

Operating SystemmacOS, Linux, or Windows (WSL2)
RAM8 GB minimum, 16 GB+ recommended
GPURecommended for local inference (Apple Silicon, NVIDIA)
Disk Space2–10 GB per model
DockerOptional — only for web search and TTS

Supported Backends

MLXOllamavLLMllama.cppOpenAI API

Frequently Asked Questions

Architecture

User

Web App (Next.js)

Chat UIVoiceVisionDocument BrainAuth
Control

Gateway API (FastAPI)

JWT AuthSSE StreamingRAG PipelineWeb SearchVoice
Thinking LLM:8080
Instant LLM:8081
SearXNG:8888
Kokoro TTS:8880
Data

SQLite + Local Storage

AuthUser DataEmbeddingsFilesFTS5
View Full Architecture — 13 Diagrams