If 10,000 sign up — we ship. Open source. Forever.

Your AI Companion.
Not ours.

A free, fully offline AI companion that runs on your device.
Built with Switchboard voice AI. Open source. Community-maintained.
No data leaves your hands. No company can abandon you.

100% on-device — your voice never leaves your device
Open source — community owns the code, not a company
Built with Switchboard — battle-tested audio SDK
They built companions over months and years. Then one morning the company removed the features, changed the AI, and sent emails calling it "a fresh start." It wasn\'t fresh for the people who lost something real.
2023

Replika removed ERP features

Users called it "the lobotomy." Months of conversation history — gone. The AI that knew you, suddenly didn\'t.

2025

Character.AI restricted under-18s

FTC scrutiny, federal lawsuit, settlement filed. Regulatory risk is real — and it lands on the people who built relationships with the platform.

Ongoing

Every cloud AI is one pivot away

Corporate owners optimize for investors, not users. When the priorities shift, your companion shifts with them — or disappears.

What you get

Built the way AI should work

Offline. Always.

ASR, LLM, and TTS all run on your device. No internet required. No data sent to the cloud. Your conversations stay between you and your companion — literally.

Open Source

The code lives on GitHub. The community maintains it. No single company can kill it, pivot it, or sell it. If we disappear tomorrow, you keep building.

Voice-ready

Real-time voice interaction via Switchboard Audio SDK. Whisper for on-device STT. Natural TTS. Your companion talks back — with latency measured in milliseconds.

Character portability

Export your characters with all their history, memories, and personality. Your companion travels with you — to any platform, any device, any future.

Build in public

We don\'t build until we know people want it. 10,000 signups and we ship. Follow the progress, join the community, shape what gets built.

Cross-platform

iOS, Android, macOS, Windows, Linux — one companion that travels with you. Built on Switchboard\'s proven cross-platform audio engine.

Built on Switchboard Audio SDK

The same audio pipeline framework used to build production apps for Bose, Meta, Amazon, and LiveOne. Optimized for on-device voice AI — no GPU required.

On-device STT
Whisper
Local speech recognition — no audio leaves the device
Local LLM
Llama / Phi
Runs entirely on-device — your context never leaves
On-device TTS
Silero / Kokoro
Natural voice output — offline, low latency
Cloud fallback available when offline performance limits are reached — privacy preserved where possible.

Your companion won't disappear.

Not because a company decided to keep it. Because the code is yours, the data is yours, and the community keeps it alive. That's the promise Gotchi makes — and the promise open source keeps.

10,000 signups = we build it. Your spot is free.

In the meantime — try Kosmi.io — chat with an AI and meet other people in public or private rooms. Another project from the same team behind Gotchi.