Local Waifu - A private AI companion that lives on your computer,

Hi Product Hunt 👋 I started building Local Waifu around one simple idea: if you spend months or years building a relationship with an AI companion, you should actually own it. Most companion apps are services. Your conversations, memory, character and even generated images depend on someone else's servers. If the pricing changes, a model disappears, a policy changes, or the service shuts down, there isn't much you can do. I wanted to build the opposite. Local Waifu is an AI companion that lives on your own computer. The language model, long-term memory, voice and image generation can all run locally. There is no account, no telemetry and no server of mine sitting between you and your character. Turn off Wi-Fi and she keeps working. And with today's launch, Local Waifu is now available on **Windows as well as Mac**. 🎉 We also added some of the biggest pieces I've wanted to bring to the project: • **Local + cloud image generation** — ask your character for a selfie or any image and she can generate it directly on your GPU. Cloud models are optional if you prefer them. • **Voice responses and full voice calls** — talk instead of typing, with speech recognition and voice generation running locally. • **Vision** — send an image in chat and your companion can actually see and respond to what's in it. The part I've spent the most time on is still memory. Conversations are stored locally, with embeddings and a knowledge graph built from the people, places and things you talk about. The goal is for the character to know you months later, not just remember the last few messages. Memory also isn't tied to one AI model. You can swap the model underneath the character and keep the same history, personality and relationship. Characters grow through seven relationship stages, have a changing mood state and can shift between different conversational modes. You build the personality yourself rather than choosing a premade character from a catalog. The app ships with its own model runtime and recommends a model based on your hardware, so you don't need Ollama, Python, a terminal or a separate AI setup. If your machine isn't powerful enough — or you simply prefer a hosted model — you can also bring your own OpenAI, Claude, Gemini, DeepSeek, Mistral, Groq, OpenRouter or compatible API key. That's completely optional and off by default. One thing that matters a lot to me is that the product isn't dependent on me continuing to run a service. Your characters and memories live on your machine. If I stopped working on Local Waifu tomorrow, the app you already have wouldn't suddenly lose its brain because a server disappeared. The tradeoffs are pretty straightforward: Local Waifu is closed-source, a good local model needs a reasonably modern computer, and a small local model won't always match the smartest cloud models. It's **$20 once**, with a **7-day free trial** and **30-day refund**. No subscription. I'd genuinely love feedback — especially from people who have used other AI companions or already run local models. And I'm happy to answer the skeptical questions too. 🙂