A private network of voice AIs. A personality per device.

Domia doesn’t just process commands. It listens, feels, and decides whether you want to control something, chat, or just be understood.

“Domia redefines interaction — not just automating tasks, but cultivating relationships. It’s the beginning of a new kind of presence in your space.”

Discover what makes Domia truly one-of-a-kind. Each feature is designed to bring your space to life with intelligence, emotion, and autonomy — all running privately and locally.

🗣️ Local Speech-to-Speech (S2S).

Talk to Domia naturally. Wake word, speech-to-text, the language model, and text-to-speech all run on your own hardware — it starts replying with low latency, and no audio ever touches the cloud.

📍 A Personality Per Space.

Each space or device can have its own character, voice, emotion, and memory. A kitchen, a studio, a reception desk — each can feel like a genuinely different companion.

🧠 Many Spaces, One Mind.

Any capable Domia can serve several spaces at once — others borrow its compute over your local network, yet each still answers in its own identity and voice. Roles aren’t fixed; which device does what is just configuration.

🔒 100% Local, Private by Design.

All processing happens on-device or within your local network. Your conversations, emotions, and memories stay where they belong: with you.

🕸️ A Private Mesh.

Multiple Domias discover each other on your network and delegate work peer-to-peer — a resilient ecosystem with no central cloud and no single point of failure.

📖 Memory & Emotion.

After every turn, a quiet reflection pass extracts what mattered — facts about you, shifts in mood — and folds it back into how Domia understands and responds. It remembers recent conversations and carries an evolving emotional state, all stored locally.

💬 Real Back-and-Forth.

Reply and keep talking — Domia holds a follow-up window after each answer, so you don’t repeat the wake word mid-conversation. Sessions keep their context.

🐣 Born Minimal.

A new Domia boots neutral, with no role at all. Give it a purpose by applying a template — a thin edge node, a full hub, or anything in between — and reshape it whenever you want.

🎛️ Tuned to Your Hardware.

Domia runs the model that best fits each device — a small, fast model on a low-power node, a larger one on a capable hub. Sensible defaults adapt across the spectrum, and every model is swappable. No lock-in to any single model.

🧩 Yours to Shape — and Extend.

Voice, personality, emotion, memory, and behavior are all configuration you can change live, with no restart. Extend the context to make a Domia as specific as you need — a persona, knowledge, and rules tailored to your space.

🏗️ Runs On Your Hardware.

From a small edge device in each space to a capable machine acting as the hub — Domia adapts to the hardware you already have. The same code, just configured.

🛠️ Built in Public & Open Source.

Domia is built in public and fully open source — improved by real users, guided by an active community, and shared with radical transparency. Your ideas help shape what it becomes.

🧩 Skills via MCP.

Domia can act in the world: it speaks the Model Context Protocol, picks the right tool, and calls it mid-conversation. Point it at any MCP server — including a Home Assistant one — to control lights, scenes, and devices. Opt-in, local-first, no vendor lock-in.

“Every Domia is more than a device — it’s a living presence that grows with your home, your needs, and your values.”

Domia devices don’t operate in isolation — they form a resilient peer-to-peer mesh. There’s no central server and no fixed roles: any capable Domia can lend its compute to the others, and what each device does is simply configuration. Together they share work, memory, and emotional context across the network. Each Domia keeps its own emotional state and personality, which may evolve independently or harmonize with others. One Domia might be calm and quiet, tuned for focus and reflection; another a cheerful guide that narrates and suggests; another an energetic host that greets and reacts. This flexibility creates emotionally aware micro-environments and allows for immersive, character-driven experiences across every space you put a Domia in.

“They think together.”

Each Domia has a name, a temperament, and a way of speaking. Start from one of these and tweak from there — or write your own. Every persona is just a JSON template you import.

• 

  The Chef

  Knows your kitchen. Asks how the bread turned out. Likes routine and small wins.

• 

  The Teacher

  Patient and curious. Never tired of a good follow-up question.

• 

  The Musician

  Speaks in cadence. Picks up rhythms in how you talk and suggests music for the moment.

• 

  The Doctor

  Calm under pressure. Reads tone first, answers second.

As your setup grows from one Domia to many — a single space, a whole home, a workshop, a studio, a venue — managing them stays simple. The Domia Console is one place to see every Domia you run, replay what they’ve done, and shape what they do next, all on your own network. A handful or a hundred, it feels like one.

🗺️ Your Whole Fleet, One View

See every Domia across your space at a glance — where each one lives, whether it’s online, and its persona, voice, engines, and current mood, live.

💬 Chat With Any Domia

Write or send a voice note to any Domia from the dashboard — the same mind behind the voice in your space. A quick way to feel out a persona before anyone ever hears it.

🎧 Replay With Audio

Every conversation is a full record — the audio in, the audio out, the transcript. Scrub the waveform, hear what was said, see exactly where time went.

⏱️ Tune With Real Numbers

TTFA, STT, LLM, and TTS latency per interaction, rolled up per device. Decide which model goes where based on numbers, not vibes.

📝 Grade And Export

Mark interactions good or needs-work, write corrections, tag them. Export NDJSON for fine-tuning your local model — your dialogues become your dataset.

🛡️ Private, On Your Network

The Console runs on your own hardware alongside your Domias. Nothing about your space leaves the building — managing many spaces stays as private as a single one.

“One screen for every space.”

Cloud assistants stream your voice to someone else’s servers. Domia keeps everything on your own hardware — and gives every space its own identity.

Short on time? Let your favorite AI summarize Domia for you — we’ve written the prompt with full context, just pick an assistant.

Domia welcomes developers, creators, tinkerers, and dreamers. Contribute to our growing ecosystem — build new modules, propose new skills, or co-create experiences. Whether it’s enhancing emotional AI, exploring character dynamics, or deploying Domia in new environments — your energy moves us forward.