Ezza

Self-hosted · Local-first · Invite-only

Your AI. Your server. Every screen.

Ezza is one self-hosted workspace for chat, a coding agent, kanban, and workflow automation — running against your own models, on your own hardware.

Explore the modes

One shell · Four modes

Not a bundle. One app.

A shared shell with a mode switcher — one session model, one settings system, from the browser down to the glasses. These previews mirror the real app UI.

Ezzav2.4
New chat
Chat
Code
Kanban
Automate
CD
Cameron
cameron@dunn…
Previous chats
llama.cpp endpoint setup
Nightly backup workflow
G2 launcher reorder
Chat · Home
Search everything⌘K
Point Ezza at my llama.cpp server and check that it responds.
E
Ezza
Thought for 6.4s · reasoning visible
httpGET /v1/models · 200web_searchbrave

Your endpoint at 192.168.1.40:8080/v1 is live. I registered it as provider “llama.cpp” and verified streaming plus tool calls. Want me to save a memory for the endpoint?

Ask Ezza anything… llama.cpp · qwen3
visible reasoningtool-activity chipsbranching historySmart SearchLive Chat · barge-in

Voice · Live Chat · Even G2

Interrupt it mid-sentence.

Live Chat is a hands-free voice loop with on-device VAD and barge-in (Android & desktop). On the Even G2, hold the temple and the same session answers on your lenses.

Live Chat

Listening…
Say something. I'll pick you up.

Speak naturally — I'll listen, respond out loud, and you can interrupt me anytime.

Push-to-talk speech-to-text, low-latency streaming TTS read-aloud, and a four-phase voice calibration. Barge-in stops the assistant the moment you start speaking. The status circle above is the app's real Live Chat panel.

on-device VADbarge-in4-phase calibrationaudio.cpp endpoints

G2 Live · Even smart glasses

EzzaOS on the Even G2 monochrome display

HOLD TEMPLE TO TALK · LC3 ↑ · STT → SAME SESSION

LC3 audio over BLE → STT → the same chat sessions → streamed reply on the glasses display. The phone proxies everything. These are the real glasses app screens — switch between them above. Launcher apps are reorderable and toggleable from phone settings.

font scaledepthbrightnesstimeout640×480 · mono

Feature parity · One codebase

The same app, everywhere you are.

One Kotlin Compose Multiplatform codebase — all business logic shared across web, desktop, and Android, plus a voice-first client for the Even G2.

Kotlin 2.4.20CMP 1.12.1
Capability WebwasmJs · Ezza server DesktopJVM · fat JAR + installer Androidnative · minSdk 24 Even G2BLE · LC3 · in Android app
Chat & shared sessions✓✓✓voice-first
Code agent✓✓✓launcher
Kanban✓✓✓launcher
Automate✓✓✓—
Voice · TTS / STT✓✓✓hold-to-talk
Live Chat · VAD loop—✓✓—
Settings & per-user sync✓✓✓via phone

Web

Kotlin/wasmJs in the browser, served by the dedicated Ezza server. Production access over HTTPS via Cloudflare Zero Trust.

Desktop

JVM app shipped as a fat JAR and Windows installer. Identical UI and features.

Android

Native app from the same shared code. Runtime permissions and keyboard-aware layout.

Even G2

Voice-first BLE client inside the Android app. The phone proxies everything; sessions stay shared.

Local-first · Self-hosted

Your hardware. Your models. Your data.

Privacy and ownership are the headline, not an afterthought. Ezza runs against your own server and your own models — nothing else is required.

A sync layer, not a platform

A lightweight Ktor server keeps per-user state in file-backed JSON — no database. Clients hold their own state and sync across devices.

Bring your own models

Any OpenAI-compatible endpoint — llama.cpp, for example — configured centrally under Providers & Models. Voice runs on audio.cpp-compatible TTS/STT endpoints.

Invite-only, multi-user

Google sign-in on every platform. Admin and member roles, one-time invite links, per-user data isolation, and profile pictures.

local-first over cloud-first unified over fragmented honesty over hype no billing no analytics one fixed dark theme

Honest UI is a contract: every surface shows real data or an explicit empty/error state — never dummy placeholders. Limitations are documented as v1 boundaries, not hidden.

Control · Depth

Governance without the enterprise.

agents · tools

Three-level tool gate

Global → per-agent → per-feature. The agent registry carries per-agent model overrides; every tool call passes the gate before it runs.

global → agent → feature
memories

Agent-managed memories

Persistent facts the assistant can create, update, and remove from chat. You review and edit them in settings — nothing is hidden.

create · update · remove
backup / restore

Full export and import

Back up the entire server state and restore it elsewhere. File-backed JSON keeps backups plain, portable, and yours.

export · import
settings

One settings system

Providers & Models, Integrations, MCP, Users, Voice, Agents, Tools, Chat, Code, Wearable Integration, Memories, Backup / Restore, Labs — the same panels on every platform.

13 panels

Run Ezza on your own server.

Deploy the Ktor sync layer, point it at your models, and invite your people with one-time links.

v1 · no billing · no analytics · one fixed dark theme

What deploying Ezza looks like

ServerLightweight Ktor sync layer. File-backed JSON, no database. Clients keep state and sync per-user across devices.
ClientsWeb (wasmJs) served by the Ezza server · Desktop JVM fat JAR + Windows installer · Android minSdk 24 · Even G2 voice client inside the Android app.
AccessGoogle sign-in on every platform. Admin/member roles, one-time invite links, per-user isolation. Production web access over HTTPS via Cloudflare Zero Trust.
ModelsAny OpenAI-compatible endpoint (e.g. llama.cpp), configured centrally. Voice on audio.cpp-compatible TTS/STT endpoints.
Everything stays on your hardware.

Search everything

⌘K
6 entries · demo history
Exact-phrase demo · the real index covers full history