Solaura AI

Solaura is a private AI workspace that brings powerful language models, a long-term memory vault, and a personal knowledge layer together — right on your device.

Designed around privacy and agency, Solaura lets you run AI where you want it: fully on-device and offline, or connected to your own cloud API. Your conversations, notes, and memories stay on your phone.

🌟 TALK TO POWERFUL AI MODELS
• Cloud Chat — Bring Your Own Key: connect any OpenAI- or Anthropic-compatible provider (OpenAI, Anthropic, DeepSeek, OpenRouter, vLLM, Ollama, LM Studio, and more). Just add your API key and base URL.
• Real-time streaming responses with rich Markdown rendering.
• Vision support — share images and let the model see them.
• Document understanding — attach PDF, DOCX, Markdown, TXT, JSON, HTML, CSV, or XLSX files. Solaura extracts the text so the model can read it.
• Tool & function calling — the assistant can use tools to search your notes and memory, propose new memories, and connect to external services.
• Save and revisit full multi-turn conversations with a built-in chat history.
• Fine-tune behavior with adjustable temperature, top-p, max tokens, and more.

🤖 RUN AI ENTIRELY ON YOUR DEVICE
• Local LLM (llama.cpp / GGUF): run open models like Llama, Qwen, Gemma, Mistral, Phi-3, DeepSeek, and Falcon completely offline. Supports vision models, reasoning/thinking models, and on-device tool use.
• Gemma on-device: run Google Gemma via LiteRT-LM, with image input, thinking, and function calling.
• Built-in OS AI: tap into your phone's built-in intelligence — Gemini Nano on supported Android devices, or Apple Foundation Models on iOS — running entirely offline.
• Pick your preferred compute backend (CPU / GPU / NPU) and tune sampling, context size, GPU layers, flash attention, and KV-cache quantization.
• Import your own model files directly from device storage.

🔎 DISCOVER & DOWNLOAD OPEN MODELS
• Neural Hub: browse thousands of trending open-source models from Hugging Face. Search, filter by format, check device compatibility, and download in a tap.
• Download Hub: a full download manager with pause/resume, retry, live progress, share, and export — engineered to handle multi-gigabyte model files safely.

🧠 A MEMORY THAT ACTUALLY REMEMBERS YOU
• Memory Space is an on-device agentic memory vault organized into Facts, Events, People, Projects, Decisions, and Insights.
• The assistant reads your memory to personalize responses, and proposes new memories — but you stay in control. Every proposed memory lands in an Inbox where you approve or reject it. Nothing is remembered without your say-so.

📝 NOTES WITH SEMANTIC AI SEARCH
• Write and organize notes with Markdown, groups, and pinning.
• Search smarter with on-device semantic search powered by a local embedding model that runs entirely offline. Your notes never leave your device.

🛠️ EXTEND THE ASSISTANT YOUR WAY
• Skills: create and organize on-device AI skills as Markdown files, with a full editor, live preview, YAML frontmatter, and categories.
• MCP (Model Context Protocol): connect Solaura to external MCP servers over SSE or HTTP to give the assistant new tools and data sources.

🔒 PRIVATE BY DESIGN
• On-device inference, on-device embeddings, and on-device storage keep your data on your phone.
• Bring your own API keys — Solaura never sells or shares your data.
• Open and transparent: review the open-source licenses and privacy policy right inside the app.

Whether you want a smarter second brain, a private offline chat assistant, or a single workspace for local and cloud AI — Solaura puts it all in your hands.

Solaura — your private AI workspace.