Liora: One Windows AI hub for chat, coding, voice and creation

Hi Product Hunt — I’m the independent developer behind SAINO, and today I’m introducing Liora. Liora is a Windows AI workspace built for people who use more than one AI. Instead of opening a separate application for local chat, coding agents, voice, search, memory and creative generation, Liora brings those workflows together in one desktop environment. I built Liora because choosing an AI should depend on the task. A local model may be the right choice for a private conversation or document search. Codex or Claude Code may be better for a development task. ComfyUI may be the right environment for image or video generation. Liora acts as the hub that connects these tools without pretending that one model is best at everything. What Liora includes: • Local AI through Ollama, LM Studio and llama.cpp • Codex and Claude Code integration through their official CLI login systems • Chat history, projects, model switching and long-term memory • Japanese-first semantic retrieval using Ruri v3 embeddings • File and document search for text, PDF, Word and knowledge sources • Web search with citations and switchable providers • Voice input, speech recognition, speech synthesis and interruption-aware conversation • Work Mode for project files, plans, tasks, Git, tests, terminals and code editing • Creative Mode for supported image, video and music generation environments • MCP servers, tools, skills, plugins, automation, backup and session recovery Liora is local-first, but not local-only. Conversations, memories, settings and logs stay in its Windows application data by default. When you intentionally enable Codex, Claude Code, web search, OAuth or another external integration, that service’s own account, privacy and pricing rules apply. Liora is designed for developers, creators and local-AI enthusiasts who are comfortable installing Windows software, downloading models and configuring tools. It is proprietary commercial software for Windows 10 and 11, available as a one-time purchase for $199 internationally or JPY 29,800 in Japan. The application and setup guides are included; models, cloud subscriptions, API usage and third-party services are not. The current installer is not code signed, so customers receive the expected filename and SHA-256 checksum for verification. Our public GitHub repository contains product documentation and screenshots, not the private application source. I’d especially appreciate feedback on the unified workflow: when you move between local chat, coding agents, voice and creative tools, what context should an AI workspace preserve for you?